BrunoSanKI-PraxisAngewandte KI für Entscheider
← Alle LLMs
LLM · DeepSeek

DeepSeek

Eine Modellfamilie als lebendes Objekt: Varianten, ausgewählte Leistungsdaten und alle verbundenen BrunoSan-Quellensignale.

518Erwähnungen
82in 30 Tagen
110Quellen
35Varianten
Im Universe erkunden →History öffnen →

Modelle

Alle bekannten Varianten dieser Familie. Ein direkter Artikel-Link springt exakt zur genannten Variante.

DeepSeek · DeepSeek

DeepSeek: DeepSeek V4.1 Flash 1017

CONFIRMED

DeepSeek V4.1 Flash 1017 is the October 17, 2026 snapshot of DeepSeek V4.1 Flash, served by DeepSeek under the `deepseek-flash` API model name. V4.1 Flash is a sparse mixture-of-experts model...

1.04858MKontext
393KMax. Output
Text · BildInput
TextOutput
$0.15Input / 1M
$0.6Output / 1M
JaTools
JaReasoning
JaStructured
Erstmals bei BrunoSan: 17. Oktober 2026HF · Lizenz mit · 763.2B Parameter7 Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
cais/hleScore44Quelle ↗Community · abgerufen 2026-10-22
Testdetails

Text-only subset, without tools.

cais/hleScore42.6Quelle ↗Community · abgerufen 2026-10-22
Testdetails

Without tools.

DeepSeek · DeepSeek

DeepSeek: DeepSeek V4.1 Flash

CONFIRMED

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...

1.04858MKontext
944KMax. Output
Text · BildInput
TextOutput
$0.3Input / 1M
$1.2Output / 1M
JaTools
JaReasoning
JaStructured
Erstmals bei BrunoSan: 10. September 2026HF · Lizenz mit · 763.2B Parameter16 Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
llamaindex/ExtractBenchScore87.11Quelle ↗Community · abgerufen 2026-10-22
Testdetails

Pipeline name: deepseek_v4_1_flash_extract_oneshot_structured_output_file (served via the DeepSeek API, thinking disabled)

cais/hleScore63.9Quelle ↗Community · abgerufen 2026-10-22
Testdetails

With tools; harness not specified in the model card.

cais/hleScore36.8Quelle ↗Community · abgerufen 2026-10-22
Idavidrein/gpqaScore90.9Quelle ↗Community · abgerufen 2026-10-22
harborframework/terminal-benchScore31.2Quelle ↗Community · abgerufen 2026-10-21
Testdetails

DeepSeek Harness, Minimal mode, 1M-token context.

Weitere 20 Ergebnisse
BenchmarkWertBeleg
harborframework/terminal-bench-2.1Score90.6Quelle ↗Community · abgerufen 2026-10-21
Testdetails

DeepSeek Harness, Minimal mode, 1M-token context.

datacurve/deep-sweScore74.2Quelle ↗Community · abgerufen 2026-10-21
Testdetails

Reported as 'DeepSWE v1.1'; official mini-swe-agent harness.

llamaindex/ParseBenchScore56.57Quelle ↗Community · abgerufen 2026-10-21
Testdetails

Pipeline name: deepseek_v4_1_flash_no_thinking_parse_with_layout (served via the DeepSeek API)

llamaindex/ExtractBenchshort94.44Quelle ↗Community · 2026-09-11
Testdetails

Pipeline name: deepseek_v4_1_flash_extract_oneshot_structured_output_file (served via the DeepSeek API, thinking disabled)

harborframework/terminal-benchterminalbench_330Quelle ↗Community · 2026-09-10
Testdetails

DeepSeek Harness, Minimal mode, 1M-token context.

Idavidrein/gpqadiamond90.9Quelle ↗Community · 2026-09-10
llamaindex/ParseBenchtext_formatting66.05Quelle ↗Community · 2026-09-10
Testdetails

Pipeline name: deepseek_v4_1_flash_no_thinking_parse_with_layout (served via the DeepSeek API)

llamaindex/ParseBenchchart20.46Quelle ↗Community · 2026-09-10
Testdetails

Pipeline name: deepseek_v4_1_flash_no_thinking_parse_with_layout (served via the DeepSeek API)

llamaindex/ParseBenchmean56.57Quelle ↗Community · 2026-09-10
Testdetails

Pipeline name: deepseek_v4_1_flash_no_thinking_parse_with_layout (served via the DeepSeek API)

harborframework/terminal-benchterminalbench_431.2Quelle ↗Community · 2026-09-10
Testdetails

DeepSeek Harness, Minimal mode, 1M-token context.

harborframework/terminal-bench-2.1terminalbench_2_190.6Quelle ↗Community · 2026-09-10
Testdetails

DeepSeek Harness, Minimal mode, 1M-token context.

datacurve/deep-swedeep_swe74.2Quelle ↗Community · 2026-09-10
Testdetails

Reported as 'DeepSWE v1.1'; official mini-swe-agent harness.

llamaindex/ExtractBenchmean87.11Quelle ↗Community · 2026-09-11
Testdetails

Pipeline name: deepseek_v4_1_flash_extract_oneshot_structured_output_file (served via the DeepSeek API, thinking disabled)

cais/hlehle36.8Quelle ↗Community · 2026-09-10
llamaindex/ExtractBenchlong22.23Quelle ↗Community · 2026-09-11
Testdetails

Pipeline name: deepseek_v4_1_flash_extract_oneshot_structured_output_file (served via the DeepSeek API, thinking disabled)

llamaindex/ParseBenchlayout28.61Quelle ↗Community · 2026-09-10
Testdetails

Pipeline name: deepseek_v4_1_flash_no_thinking_parse_with_layout (served via the DeepSeek API)

llamaindex/ParseBenchtable79.65Quelle ↗Community · 2026-09-10
Testdetails

Pipeline name: deepseek_v4_1_flash_no_thinking_parse_with_layout (served via the DeepSeek API)

llamaindex/ExtractBenchmedium81.49Quelle ↗Community · 2026-09-11
Testdetails

Pipeline name: deepseek_v4_1_flash_extract_oneshot_structured_output_file (served via the DeepSeek API, thinking disabled)

cais/hlehle63.9Quelle ↗Community · 2026-09-10
Testdetails

With tools; harness not specified in the model card.

llamaindex/ParseBenchtext_content88.09Quelle ↗Community · 2026-09-10
Testdetails

Pipeline name: deepseek_v4_1_flash_no_thinking_parse_with_layout (served via the DeepSeek API)

DeepSeek · DeepSeek

DeepSeek: DeepSeek V4.1 Flash (batch)

CONFIRMED

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...

1.04858MKontext
384KMax. Output
Text · BildInput
TextOutput
$0.05Input / 1M
$0.15Output / 1M
JaTools
JaReasoning
JaStructured
Erwähnungen →Universe →Bei OpenRouter ansehen ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek: DeepSeek V4 Flash Vision Exp

CONFIRMED

DeepSeek V4 Flash Vision Exp is an experimental vision-enabled version of [DeepSeek V4 Flash 0731](https://openrouter.ai/deepseek/deepseek-v4-flash-0731) from DeepSeek, adding image understanding while matching the base model on text capabilities including agents,...

1.04858MKontext
262KMax. Output
Text · BildInput
TextOutput
$0.2156Input / 1M
$0.6468Output / 1M
JaTools
JaReasoning
JaStructured
Erstmals bei BrunoSan: 21. August 20261 Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
hkust-nlp/ToolathlonScore75.9Quelle ↗Community · abgerufen 2026-10-22
harborframework/terminal-bench-2.1Score83.9Quelle ↗abgerufen 2026-10-21
datacurve/deep-sweScore59.3Quelle ↗abgerufen 2026-10-21
DeepSeek · DeepSeek

DeepSeek: DeepSeek V4 Pro 0813

CONFIRMED

DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts model from DeepSeek. This is the GA release of DeepSeek V4 Pro.

1.04858MKontext
393KMax. Output
TextInput
TextOutput
$0.66Input / 1M
$1.98Output / 1M
JaTools
JaReasoning
JaStructured
Erstmals bei BrunoSan: 13. August 20264 Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
hkust-nlp/ToolathlonScore74.1Quelle ↗Community · abgerufen 2026-10-22
harborframework/terminal-bench-2.1Score87.9Quelle ↗Community · abgerufen 2026-10-21
Testdetails

DeepSeek Harness (minimal mode), max reasoning effort, temperature=1.0, top_p=0.95.

DeepSeek · DeepSeek

DeepSeek: DeepSeek V4 Flash 0731

CONFIRMED

DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....

1.04858MKontext
131KMax. Output
TextInput
TextOutput
$0.0444Input / 1M
$0.1333Output / 1M
JaTools
JaReasoning
JaStructured
Erstmals bei BrunoSan: 31. Juli 2026HF · Lizenz mit · 304.2B Parameter6 Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
hkust-nlp/ToolathlonScore70.3Quelle ↗Community · abgerufen 2026-10-22
harborframework/terminal-bench-2.1Score82.7Quelle ↗Community · abgerufen 2026-10-21
Testdetails

DeepSeek Harness (minimal mode), max reasoning effort, temperature=1.0, top_p=0.95.

datacurve/deep-sweScore54.4Quelle ↗Community · abgerufen 2026-10-21
Agents' Last ExamScore25.2Quelle ↗Model Card · abgerufen 2026-10-10
NL2RepoScore54.2Quelle ↗Model Card · abgerufen 2026-10-10
Weitere 10 Ergebnisse
BenchmarkWertBeleg
DSBench-FullStack †Score68.7Quelle ↗Model Card · abgerufen 2026-10-10
Terminal Bench 2.1Score82.7Quelle ↗Model Card · abgerufen 2026-10-10
DeepSWEScore54.4Quelle ↗Model Card · abgerufen 2026-10-10
DSBench-Hard †Score59.6Quelle ↗Model Card · abgerufen 2026-10-10
AutomationBench PublicScore25.1Quelle ↗Model Card · abgerufen 2026-10-10
Toolathlon-VerifiedScore70.3Quelle ↗Model Card · abgerufen 2026-10-10
CybergymScore76.7Quelle ↗Model Card · abgerufen 2026-10-10
harborframework/terminal-bench-2.1terminalbench_2_182.7Quelle ↗Community · 2026-08-01
Testdetails

DeepSeek Harness (minimal mode), max reasoning effort, temperature=1.0, top_p=0.95.

datacurve/deep-swedeep_swe54.4Quelle ↗Community · 2026-08-03
hkust-nlp/Toolathlontoolathlon_verified70.3Quelle ↗Community · 2026-08-01
DeepSeek · DeepSeek

DeepSeek: DeepSeek V4 Pro 0423

CONFIRMED

DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...

1.04858MKontext
384KMax. Output
TextInput
TextOutput
$0.4427Input / 1M
$0.8853Output / 1M
JaTools
JaReasoning
JaStructured
Erwähnungen →Universe →Bei OpenRouter ansehen ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek: DeepSeek V4 Flash 0423

CONFIRMED

DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...

1.04858MKontext
944KMax. Output
TextInput
TextOutput
$0.091Input / 1M
$0.182Output / 1M
JaTools
JaReasoning
JaStructured
Erwähnungen →Universe →Bei OpenRouter ansehen ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek: DeepSeek V3.2

CONFIRMED

DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...

164KKontext
66KMax. Output
TextInput
TextOutput
$0.28Input / 1M
$0.42Output / 1M
JaTools
JaReasoning
JaStructured
HF · Lizenz mit · 685.4B Parameter Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
open-agent-leaderboard/resultsoverall0.4585Quelle ↗Community · 2026-05-18
Testdetails

agent: React

open-agent-leaderboard/resultstaubench_retail0.77Quelle ↗Community · 2026-05-18
Testdetails

agent: Smolagent

open-agent-leaderboard/resultsbrowsecomp_plus0.48Quelle ↗Community · 2026-05-18
Testdetails

agent: Claude Code

open-agent-leaderboard/resultsswebench0.64Quelle ↗Community · 2026-05-18
Testdetails

agent: Claude Code

TIGER-Lab/MMLU-Prommlu_pro85Quelle ↗Community · 2026-02-28
Weitere 47 Ergebnisse
BenchmarkWertBeleg
open-agent-leaderboard/resultsappworld0.09Quelle ↗Community · 2026-05-18
Testdetails

agent: React

mercor/apex-agentsapex-agents7Quelle ↗Community · 2026-04-22
Testdetails

harness: react_toolbelt_agent; reasoning_effort: high

open-agent-leaderboard/resultstaubench_telecom0.18Quelle ↗Community · 2026-05-18
Testdetails

agent: OpenAI Solo

open-agent-leaderboard/resultsoverall0.4158Quelle ↗Community · 2026-05-18
Testdetails

agent: Claude Code

Idavidrein/gpqadiamond82.4Quelle ↗Community · 2026-02-28
open-agent-leaderboard/resultstaubench_retail0.82Quelle ↗Community · 2026-05-18
Testdetails

agent: React + Shortlisting

open-agent-leaderboard/resultstaubench_telecom0.84Quelle ↗Community · 2026-05-18
Testdetails

agent: Smolagent

claw-eval/Claw-Evalgeneral42.2Quelle ↗Community · 2026-04-23
Testdetails

Pass³% | N=3 | 161 tasks

open-agent-leaderboard/resultstaubench_retail0.82Quelle ↗Community · 2026-05-18
Testdetails

agent: React

open-agent-leaderboard/resultsoverall0.3217Quelle ↗Community · 2026-05-18
Testdetails

agent: OpenAI Solo

open-agent-leaderboard/resultsswebench0.6875Quelle ↗Community · 2026-05-18
Testdetails

agent: React + Shortlisting

internlm/WildClawBenchoverall34Quelle ↗Community · 2026-05-22
open-agent-leaderboard/resultstaubench_telecom0.71Quelle ↗Community · 2026-05-18
Testdetails

agent: React

open-agent-leaderboard/resultsoverall0.446Quelle ↗Community · 2026-05-18
Testdetails

agent: React + Shortlisting

open-agent-leaderboard/resultsbrowsecomp_plus0.3Quelle ↗Community · 2026-05-18
Testdetails

agent: OpenAI Solo

cais/hlehle40.8Quelle ↗Community · 2026-02-13
Testdetails

No tools, text-only

open-agent-leaderboard/resultsappworld0.06Quelle ↗Community · 2026-05-18
Testdetails

agent: OpenAI Solo

harborframework/terminal-bench-2.0terminalbench_239.6Quelle ↗Community · 2026-02-10
Testdetails

agent: Terminus 2

MathArena/hmmt_feb_2026MathArena/hmmt_feb_202684.09Quelle ↗Community · 2026-02-23
open-agent-leaderboard/resultsoverall0.4092Quelle ↗Community · 2026-05-18
Testdetails

agent: Smolagent

open-agent-leaderboard/resultsbrowsecomp_plus0.21Quelle ↗Community · 2026-05-18
Testdetails

agent: Smolagent

open-agent-leaderboard/resultstaubench_telecom0.71Quelle ↗Community · 2026-05-18
Testdetails

agent: React + Shortlisting

open-agent-leaderboard/resultsswebench0.6875Quelle ↗Community · 2026-05-18
Testdetails

agent: React

internlm/WildClawBenchavg_time549Quelle ↗Community · 2026-05-22
SWE-bench/SWE-bench_Verifiedswe_bench_%_resolved70Quelle ↗Community · 2026-02-28
Testdetails

high reasoning

open-agent-leaderboard/resultstaubench_airline0.56Quelle ↗Community · 2026-05-18
Testdetails

agent: React

open-agent-leaderboard/resultstaubench_telecom0.61Quelle ↗Community · 2026-05-18
Testdetails

agent: Claude Code

MathArena/aime_2026MathArena/aime_202694.17Quelle ↗Community · 2026-02-18
collinear-ai/yc-benchmedium125263Quelle ↗Community · 2026-04-02
Testdetails

medium preset, avg final funds (USD) across seeds 1,2,3

open-agent-leaderboard/resultstaubench_retail0.65Quelle ↗Community · 2026-05-18
Testdetails

agent: Claude Code

open-agent-leaderboard/resultstaubench_airline0.2Quelle ↗Community · 2026-05-18
Testdetails

agent: OpenAI Solo

open-agent-leaderboard/resultsswebench0.56Quelle ↗Community · 2026-05-18
Testdetails

agent: Smolagent

open-agent-leaderboard/resultstaubench_airline0.56Quelle ↗Community · 2026-05-18
Testdetails

agent: React + Shortlisting

open-agent-leaderboard/resultsbrowsecomp_plus0.36Quelle ↗Community · 2026-05-18
Testdetails

agent: React + Shortlisting

internlm/WildClawBenchavg_cost11.4Quelle ↗Community · 2026-05-22
ScaleAI/SWE-bench_ProSWE_Bench_Pro15.56Quelle ↗Community · 2026-02-28
open-agent-leaderboard/resultstaubench_airline0.6Quelle ↗Community · 2026-05-18
Testdetails

agent: Smolagent

mercor/APEX-SWEapex-swe15.8Quelle ↗Community · 2026-04-22
Testdetails

harness: claude_code; reasoning_effort: high

open-agent-leaderboard/resultsswebench0.7368Quelle ↗Community · 2026-05-18
Testdetails

agent: OpenAI Solo

open-agent-leaderboard/resultstaubench_airline0.28Quelle ↗Community · 2026-05-18
Testdetails

agent: Claude Code

FutureMa/EvasionBenchevasion_bench66.88Quelle ↗Community · 2026-02-10
claw-eval/Claw-Evalmulti_turn31.6Quelle ↗Community · 2026-04-23
Testdetails

Pass³% | N=3 | 38 tasks

open-agent-leaderboard/resultsappworld0.04Quelle ↗Community · 2026-05-18
Testdetails

agent: React + Shortlisting

open-agent-leaderboard/resultsbrowsecomp_plus0.36Quelle ↗Community · 2026-05-18
Testdetails

agent: React

open-agent-leaderboard/resultsappworld0.13Quelle ↗Community · 2026-05-18
Testdetails

agent: Smolagent

open-agent-leaderboard/resultstaubench_retail0.19Quelle ↗Community · 2026-05-18
Testdetails

agent: OpenAI Solo

open-agent-leaderboard/resultsappworld0.03Quelle ↗Community · 2026-05-18
Testdetails

agent: Claude Code

DeepSeek · DeepSeek

DeepSeek: DeepSeek V3.2 Exp

CONFIRMED

DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...

164KKontext
147KMax. Output
TextInput
TextOutput
$0.27Input / 1M
$0.41Output / 1M
JaTools
JaReasoning
JaStructured
Erstmals bei BrunoSan: 30. September 2025HF · Lizenz mit · 685.4B Parameter2 Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
MMLU-ProScore85.0Quelle ↗Model Card · abgerufen 2026-10-14
SWE VerifiedScore67.8Quelle ↗Model Card · abgerufen 2026-10-14
GPQA-DiamondScore79.9Quelle ↗Model Card · abgerufen 2026-10-14
Humanity's Last ExamScore19.8Quelle ↗Model Card · abgerufen 2026-10-14
CodeforcesScore2121Quelle ↗Model Card · abgerufen 2026-10-14
Weitere 10 Ergebnisse
BenchmarkWertBeleg
HMMT 2025Score83.6Quelle ↗Model Card · abgerufen 2026-10-14
SWE-bench MultilingualScore57.9Quelle ↗Model Card · abgerufen 2026-10-14
AIME 2025Score89.3Quelle ↗Model Card · abgerufen 2026-10-14
LiveCodeBenchScore74.1Quelle ↗Model Card · abgerufen 2026-10-14
Aider-PolyglotScore74.5Quelle ↗Model Card · abgerufen 2026-10-14
BrowseComp-zhScore47.9Quelle ↗Model Card · abgerufen 2026-10-14
BrowseCompScore40.1Quelle ↗Model Card · abgerufen 2026-10-14
Terminal-benchScore37.7Quelle ↗Model Card · abgerufen 2026-10-14
SimpleQAScore97.1Quelle ↗Model Card · abgerufen 2026-10-14
LEXam-Benchmark/LEXamopen_question56.53Quelle ↗Community · 2026-06-02
Testdetails

reasoning

DeepSeek · DeepSeek

DeepSeek: DeepSeek V3.1 Terminus

CONFIRMED

DeepSeek-V3.1 Terminus is an update to [DeepSeek V3.1](/deepseek/deepseek-chat-v3.1) that maintains the model's original capabilities while addressing issues reported by users, including language consistency and agent capabilities, further optimizing the model's...

164KKontext
147KMax. Output
TextInput
TextOutput
$0.27Input / 1M
$1Output / 1M
JaTools
JaReasoning
JaStructured
HF · Lizenz mit · 684.5B Parameter Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek: DeepSeek V3.1

CONFIRMED

DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...

164KKontext
33KMax. Output
TextInput
TextOutput
$0.25Input / 1M
$0.95Output / 1M
JaTools
JaReasoning
JaStructured
HF · Lizenz mit · 684.5B Parameter Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek: DeepSeek V3 0324

CONFIRMED

DeepSeek V3, a 685B-parameter, mixture-of-experts model, is the latest iteration of the flagship chat model family from the DeepSeek team. It succeeds the [DeepSeek V3](/deepseek/deepseek-chat-v3) model and performs really well...

164KKontext
115KMax. Output
TextInput
TextOutput
$0.29Input / 1M
$1.14Output / 1M
JaTools
NeinReasoning
JaStructured
Erstmals bei BrunoSan: 25. März 20253 Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
TIGER-Lab/MMLU-ProScore81.2Quelle ↗Community · abgerufen 2026-10-22
DeepSeek · DeepSeek

DeepSeek: DeepSeek V3

CONFIRMED

DeepSeek-V3 is the latest model from the DeepSeek team, building upon the instruction following and coding abilities of the previous versions. Pre-trained on nearly 15 trillion tokens, the reported evaluations...

164KKontext
16KMax. Output
TextInput
TextOutput
$0.2574Input / 1M
$1.0287Output / 1M
JaTools
NeinReasoning
JaStructured
HF · 684.5B Parameter Erwähnungen →Universe →Hugging Face ↗Bei OpenRouter ansehen ↗

Benchmarks

BenchmarkWertBeleg
LEXam-Benchmark/LEXamScore46.57Quelle ↗Community · abgerufen 2026-10-22
openai/gsm8kScore89.3Quelle ↗Community · abgerufen 2026-10-22
Idavidrein/gpqaScore58.2070707071Quelle ↗Community · abgerufen 2026-10-22
Testdetails

GPQA Diamond

TIGER-Lab/MMLU-ProScore64.4Quelle ↗Community · abgerufen 2026-10-22
DeepSeek · DeepSeek

DeepSeek

EMERGING

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erstmals bei BrunoSan: 20. November 202452 Erwähnungen →Universe →
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-V4

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erstmals bei BrunoSan: 21. Januar 202673 Erwähnungen →Universe →
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-V4-Pro

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erstmals bei BrunoSan: 24. April 2026HF · Lizenz mit · 1598.8B Parameter22 Erwähnungen →Universe →Hugging Face ↗

Benchmarks

BenchmarkWertBeleg
openai/gsm8kScore92.6Quelle ↗Community · abgerufen 2026-10-22
SWE-bench/SWE-bench_VerifiedScore80.6Quelle ↗Community · abgerufen 2026-10-22
cais/hleScore37.7Quelle ↗Community · abgerufen 2026-10-22
Idavidrein/gpqaScore90.1Quelle ↗Community · abgerufen 2026-10-22
TIGER-Lab/MMLU-ProScore87.5Quelle ↗Community · abgerufen 2026-10-22
Weitere 14 Ergebnisse
BenchmarkWertBeleg
benchflow/skillsbenchScore50.1Quelle ↗Community · abgerufen 2026-10-22
Testdetails

with-skills; BenchFlow harness; OpenHands agent; 87 tasks x 3 trials; full 261/261 coverage

actava/chi-benchScore14.2Quelle ↗Community · abgerufen 2026-10-22
Testdetails

Harness: OpenAI Agents SDK; Protocol: 75 tasks x 3 trials; Metric: pass@1 (%)

InternScience/ResearchClawBenchScore17.12Quelle ↗Community · abgerufen 2026-10-22
Testdetails

ResearchHarness evaluation with tools enabled, code execution, and a file-system workspace; completed 39/40 ResearchClawBench tasks.

claw-eval/Claw-EvalScore58.4Quelle ↗Community · abgerufen 2026-10-22
Testdetails

Pass³% | N=3 | 161 tasks

internlm/WildClawBenchScore12Quelle ↗Community · abgerufen 2026-10-22
internlm/WildClawBenchScore605Quelle ↗Community · abgerufen 2026-10-22
internlm/WildClawBenchScore43.7Quelle ↗Community · abgerufen 2026-10-22
collinear-ai/yc-benchScore1066426Quelle ↗Community · abgerufen 2026-10-22
Testdetails

medium preset, avg final funds (USD) across seeds 1,2,3

harborframework/terminal-bench-2.0Score67.9Quelle ↗Community · abgerufen 2026-10-22
IntelligenceLab/Long-Horizon-Terminal-BenchScore3Quelle ↗Community · abgerufen 2026-10-21
Testdetails

3/46 tasks solved at reward >= 0.95 (pass@0.95); mean reward x100 = 30.7; official LHTB Harbor harness

agents-last-exam/agents-last-examScore12.390350877192983Quelle ↗Community · abgerufen 2026-10-21
Testdetails

harness=openclaw; harness_variant=not reported; attempted=146/152; official leaderboard snapshot=2026-10-01T20:39:17.826877+00:00; values are percentages (0-100); equal-task aggregation; unattempted tasks contribute zero; pass threshold score>=0.999. This is an official leaderboard export, not an HF-verified evaluation.

SWE-bench/SWE-bench_MultilingualScore76.2Quelle ↗Community · abgerufen 2026-10-21
ScaleAI/SWE-bench_ProScore55.4Quelle ↗Community · abgerufen 2026-10-21
joelniklaus/LEXam-hardScore38.93Quelle ↗Community · abgerufen 2026-10-21
Testdetails

lighteval, LEXam paper prompts, one response per question, no tools; DeepSeek-R1-0528 judge; mean of the German and English means over the 518 questions, 0-100

DeepSeek · DeepSeek

DeepSeek-V3

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erstmals bei BrunoSan: 27. Dezember 2024HF · 684.5B Parameter14 Erwähnungen →Universe →Hugging Face ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-V3.2

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erstmals bei BrunoSan: 05. Dezember 2025HF · Lizenz mit · 685.4B Parameter13 Erwähnungen →Universe →Hugging Face ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-V3.1

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erstmals bei BrunoSan: 27. August 2025HF · Lizenz mit · 684.5B Parameter4 Erwähnungen →Universe →Hugging Face ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-V3.1-Terminus

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
HF · Lizenz mit · 684.5B Parameter Erwähnungen →Universe →Hugging Face ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-V2.5

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
HF · Lizenz other · 235.7B Parameter Erwähnungen →Universe →Hugging Face ↗

Benchmarks

BenchmarkWertBeleg
TIGER-Lab/MMLU-ProScore65.83Quelle ↗Community · abgerufen 2026-10-22
DeepSeek · DeepSeek

DeepSeek-V2-Lite

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
HF · Lizenz other · 15.7B Parameter Erwähnungen →Universe →Hugging Face ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-V2

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
HF · Lizenz other · 235.7B Parameter Erwähnungen →Universe →Hugging Face ↗

Benchmarks

BenchmarkWertBeleg
openai/gsm8kScore79.2Quelle ↗Community · abgerufen 2026-10-22
DeepSeek · DeepSeek

DeepSeek-MoE

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erwähnungen →Universe →
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-LLM-7B

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erwähnungen →Universe →
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-LLM-67B

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erwähnungen →Universe →
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

DeepSeek-OCR

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
HF · Lizenz mit · 3.3B Parameter Erwähnungen →Universe →Hugging Face ↗

Benchmarks

BenchmarkWertBeleg
PaddlePaddle/Real5-OmniDocBenchScore73.99Quelle ↗Community · abgerufen 2026-10-21
allenai/olmOCR-benchScore75.7Quelle ↗Community · abgerufen 2026-10-21
Delores-Lin/MDPBenchScore51.8Quelle ↗Community · abgerufen 2026-10-21
Delores-Lin/MDPBenchit69.3Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchprivate54.5Quelle ↗Community · 2026-07-03
Weitere 34 Ergebnisse
BenchmarkWertBeleg
Delores-Lin/MDPBenchjp49.1Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchdigital80.7Quelle ↗Community · 2026-07-03
allenai/olmOCR-benchmulti_column66.4Quelle ↗Community · 2026-02-19
Delores-Lin/MDPBenchoverall51.8Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchzh59.7Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchar56.9Quelle ↗Community · 2026-07-03
allenai/olmOCR-benchtable_tests80.2Quelle ↗Community · 2026-02-19
allenai/olmOCR-benchoverall75.7Quelle ↗Community · 2026-02-19
PaddlePaddle/Real5-OmniDocBenchillumination78.1Quelle ↗Community · 2026-08-08
Delores-Lin/MDPBenchhi52.2Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchfr43.2Quelle ↗Community · 2026-07-03
PaddlePaddle/Real5-OmniDocBenchoverall73.99Quelle ↗Community · 2026-08-08
allenai/olmOCR-bencharxiv_math77.2Quelle ↗Community · 2026-02-19
Delores-Lin/MDPBenchen58.3Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchvi54.1Quelle ↗Community · 2026-07-03
allenai/olmOCR-benchold_scans_math73.6Quelle ↗Community · 2026-02-19
Delores-Lin/MDPBenchphotographed42.2Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchpt53Quelle ↗Community · 2026-07-03
PaddlePaddle/Real5-OmniDocBenchskew63.01Quelle ↗Community · 2026-08-08
PaddlePaddle/Real5-OmniDocBenchscanning86.17Quelle ↗Community · 2026-08-08
Delores-Lin/MDPBenchnl52.4Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchnon_latin48.9Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchde55Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchid60.9Quelle ↗Community · 2026-07-03
allenai/olmOCR-benchold_scans33.3Quelle ↗Community · 2026-02-19
Delores-Lin/MDPBenches44.1Quelle ↗Community · 2026-07-03
PaddlePaddle/Real5-OmniDocBenchwarping67.2Quelle ↗Community · 2026-08-08
allenai/olmOCR-benchheaders_footers96.1Quelle ↗Community · 2026-02-19
Delores-Lin/MDPBenchth49.4Quelle ↗Community · 2026-07-03
allenai/olmOCR-benchlong_tiny_text79.4Quelle ↗Community · 2026-02-19
Delores-Lin/MDPBenchru36.2Quelle ↗Community · 2026-07-03
PaddlePaddle/Real5-OmniDocBenchscreen_photography75.31Quelle ↗Community · 2026-08-08
Delores-Lin/MDPBenchko28.2Quelle ↗Community · 2026-07-03
Delores-Lin/MDPBenchzh_t59.2Quelle ↗Community · 2026-07-03
DeepSeek · DeepSeek

DeepSeek-Coder-V2-Lite-Base

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erwähnungen →Universe →Hugging Face ↗

Benchmarks

BenchmarkWertBeleg
TIGER-Lab/MMLU-ProScore34.37Quelle ↗Community · abgerufen 2026-10-22
DeepSeek · DeepSeek

DeepSeek-V4-Flash

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erwähnungen →Universe →Hugging Face ↗

Benchmarks

BenchmarkWertBeleg
SWE-bench/SWE-bench_VerifiedScore79Quelle ↗Community · abgerufen 2026-10-22
cais/hleScore34.8Quelle ↗Community · abgerufen 2026-10-22
Idavidrein/gpqaScore88.1Quelle ↗Community · abgerufen 2026-10-22
TIGER-Lab/MMLU-ProScore86.4Quelle ↗Community · abgerufen 2026-10-22
benchflow/skillsbenchScore44.7Quelle ↗Community · abgerufen 2026-10-22
Testdetails

with-skills; BenchFlow harness; OpenHands agent; 87 tasks x 3 trials; full 261/261 coverage

Weitere 5 Ergebnisse
BenchmarkWertBeleg
claw-eval/Claw-EvalScore57.8Quelle ↗Community · abgerufen 2026-10-22
Testdetails

Pass³% | N=3 | 161 tasks

harborframework/terminal-bench-2.0Score56.9Quelle ↗Community · abgerufen 2026-10-22
IntelligenceLab/Long-Horizon-Terminal-BenchScore2Quelle ↗Community · abgerufen 2026-10-21
Testdetails

2/46 tasks solved at reward >= 0.95, estimated at the leaderboard's 90-minute budget from interim verifier checkpoints of a 3-hour terminus-2 run; NOT a real 90-minute run. Range 2-3/46 (riscv-core-debug unresolved). The same run scores 5/46 at its full 3-hour budget. Official LHTB Harbor harness.

SWE-bench/SWE-bench_MultilingualScore73.3Quelle ↗Community · abgerufen 2026-10-21
joelniklaus/LEXam-hardScore38.19Quelle ↗Community · abgerufen 2026-10-21
Testdetails

lighteval, LEXam paper prompts, one response per question, no tools; DeepSeek-R1-0528 judge; mean of the German and English means over the 518 questions, 0-100

DeepSeek · DeepSeek

DeepSeek-V2-Chat

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erwähnungen →Universe →Hugging Face ↗

Benchmarks

BenchmarkWertBeleg
TIGER-Lab/MMLU-ProScore54.81Quelle ↗Community · abgerufen 2026-10-22
DeepSeek · DeepSeek

DeepSeek-OCR-2

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
Erwähnungen →Universe →Hugging Face ↗

Benchmarks

BenchmarkWertBeleg
PaddlePaddle/Real5-OmniDocBenchScore73.01Quelle ↗Community · abgerufen 2026-10-21
llamaindex/ParseBenchScore41.2Quelle ↗Community · abgerufen 2026-10-21
Testdetails

Pipeline name: deepseekocr2_vllm

allenai/olmOCR-benchScore76.3Quelle ↗Community · abgerufen 2026-10-21
DeepSeek · DeepSeek

DeepSeek-V3.2-Exp-Base

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
HF · Lizenz mit · 685.4B Parameter Erwähnungen →Universe →Hugging Face ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

deepseek-vl-1.3b-chat

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
HF · Lizenz other · 2B Parameter Erwähnungen →Universe →Hugging Face ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.
DeepSeek · DeepSeek

deepseek-vl-7b-chat

KNOWN

Technische Leistungsdaten sind für diese Variante noch nicht vollständig bestätigt.

—Kontext
—Max. Output
—Input
—Output
—Input / 1M
—Output / 1M
—Tools
—Reasoning
—Structured
HF · Lizenz other · 7.3B Parameter Erwähnungen →Universe →Hugging Face ↗
Benchmarks werden ergänzt, sobald Werte vorliegen.

BrunoSan Chronik

Alle verbundenen Artikel, neueste zuerst.

22. Oktober 2026

DeepSeek V4.1-Flash-1017 vs. Claude Haiku 5.5: More accurate at a fraction of the pricethenewstack.io
Sources: DeepSeek raised ~$15B in its latest fundraising round, one of China's largest ever, with CATL and Tencent among the biggest contributors (Lulu Yilun Chen/Bloomberg)techmeme.comIM MAGAZIN

21. Oktober 2026

The Huawei-DeepSeek Symbiosisnewsletter.semianalysis.com
Deepseek: Dieses quelloffene Tool soll KI-Kosten reduzieren – für wen es gedacht istt3n.de
China protests the US Deepseek ban while blocking ChatGPT, Claude, and Gemini at homethe-decoder.com
China criticizes the US DHS's DeepSeek ban, vowing "resolute countermeasures" while rejecting allegations from the FBI and AI labs of illicit distillation (Ephrat Livni/New York Times)techmeme.com
DeepSeek Wants Better AI at Lower Costtechnode.com
U.S. Orders DeepSeek Removed From Federal Civilian Systems, Bans New Purchasesthehackernews.com
DeepSeek open-sources inference framework with support for Nvidia and Huawei chipstechnode.com
US bans DeepSeek across federal agencies, putting AI supply chains under scrutinyinfoworld.com

20. Oktober 2026

DeepSeek open-sources DeepInfer, the inference framework behind its production servicessiliconangle.com
Deepseek open-sources the inference engine that powers its fastest and cheapest AI modelthe-decoder.com
DeepSeek AI tools banned from US government agency systems as rift with China growstechradar.com
Trump Administration Bans DeepSeek Across Civilian Agencies as Chinese Startup Open-Sources DeepInfer Frameworktechstrong.ai
DeepSeek Open-Sources DeepInfer, Its In-House Inference Frameworkofficechai.com
DeepSeek open-sources the inference engine behind its API. And Huawei gets equal billing with Nvidia.thenewstack.io
DeepSeek releases DeepInfer inference framework as open sourcedataphoenix.info
DHS bans DeepSeek products across US federal civilian agenciesbleepingcomputer.com
US orders federal agencies to remove China’s DeepSeek AI from their systemstechcrunch.com
DeepSeek open-sources DeepInfer, its internal production stack designed to accelerate inference, optimize chip performance, and reduce energy and compute costs (Saritha Rai/Bloomberg)techmeme.com
US bans use of China’s DeepSeek AI tools by government agenciestheguardian.com
US bans Deepseek technology across federal civilian agenciesthe-decoder.com
US plans to remove DeepSeek from government agency systemssiliconrepublic.com
Trump administration orders ban on DeepSeek in federal agenciessiliconangle.com

19. Oktober 2026

Trump Admin Bans DeepSeek From Federal Agenciesgizmodo.com
US Government Orders DeepSeek Removed From All Federal Civilian IT Systemsofficechai.com
DeepSeek v4.1 Flash 1017 Places Ahead Of GLM 5.3, Kimi K3 On Artificial Analysis Intelligence Indexofficechai.com
MiniMax M3.1 Flash shows China's AI price race has no floor in sightthe-decoder.com
US bans DeepSeek AI from government use as tech tension with China continuesbbc.co.ukIM MAGAZIN
Sources: DeepSeek and Huawei are collaborating on future AI chip designs; DeepSeek plans to start using Ascend 950DT for pre-training as soon as 2026's end (The Information)techmeme.com
DeepSeek Opens Its In-House Search Engine To Developerssearchenginejournal.com
The US DHS orders all federal civilian agencies to remove any DeepSeek tech from their IT systems and prohibits future procurement of its products and services (Connor Hart/Wall Street Journal)techmeme.com
DeepSeek made its name with cheap AI models. Now it’s undercutting search API rivals, too.thenewstack.io
DeepSeek updates V4.1-Flash model to improve performancesiliconrepublic.com
😺 DeepSeek V4.1-Flash-1017: Open Model, Cheap Search APItheneurondaily.com
DeepSeek releases V4.1 Flash-1017 with stronger coding and agent capabilitiestechnode.com
DeepSeek launches web search and page-fetching APIs for AI agentstechnode.com

18. Oktober 2026

DeepSeek releases updated V4.1-Flash model it says completes tasks fastersiliconangle.com
DeepSeek Unveils Leaner V4.1-Flash Updatetechstrong.ai

17. Oktober 2026

DeepSeek releases V4.1-Flash-1017, an update it says offers "stronger benchmark performance, faster task completion, and shorter outputs" than its predecessor (Saritha Rai/Bloomberg)techmeme.com
Why Britain should reject the lure of cheap – and scary – Chinese AItheguardian.com
DeepSeek V4.1 Flash is cheap, fast and visual | StartupHub.aistartuphub.ai
Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task coststhe-decoder.com
DeepSeek Launches V4.1-Flash-1017, Beats Claude Opus 5 On Some Benchmarksofficechai.comIM MAGAZIN

16. Oktober 2026

DeepSeek V4.1 Flash is now available as a fully managed API in Model Gardendiscuss.google.devIM MAGAZIN
A look at China's AI learning frenzy, as economic anxiety and fear of being left behind fuel a booming market for courses promising prosperity through AI (Afra Wang/New York Magazine)techmeme.com
Microsoft unveils MAI-Base-2, a MoE model with 2.8T total and 80B active parameters, claiming 4x-12x more pretraining FLOPs-efficiency than DeepSeek-V4-Pro-Base (Microsoft AI)techmeme.com

15. Oktober 2026

DeepSeek offers user refunds after suffering another major outagescmp.com
DeepSeek implements fix after partial outage affects chat and API servicestechnode.com

14. Oktober 2026

DeepSeek’s privacy policy: what it means for your datawitness.ai

12. Oktober 2026

Sources: DeepSeek agrees to lead robotic AI startup Winverse AI's seed funding round of tens of millions of dollars, valuing the startup at ~$400M post-money (Bloomberg)techmeme.com
DeepSeek’s cost-saving AI may lead to inaccuracy in model performance: research | South China Morning Postscmp.com

10. Oktober 2026

Sources: Nvidia is in talks to acquire or invest more into US open-weights AI startup Reflection AI; the deal may be an acquihire to avoid antitrust scrutiny (Financial Times)techmeme.com

09. Oktober 2026

Mistral’s new Large 4 trails some Chinese open models in independent teststomshardware.com
ByteDance researchers identify cause of inconsistent long-context retrieval in DeepSeek modelstechnode.com

08. Oktober 2026

RTX Spark Laptops Pave The Way To Running 284B AI Models Locally That Outperform GPT-5 In Coding And Reasoning Benchmarks, According To Microsoft Executivewccftech.com
AI-powered hacking tools enabled a likely single attacker to breach multiple South Korean banksthe-decoder.comIM MAGAZIN
DeepSeek reportedly nears $12 billion funding round backed by Tencent and CATLtechnode.com

07. Oktober 2026

Huawei and DeepSeek release open source AI tools in a bid to lower Nvidia exposure — but will programmers make the switch?techradar.com

06. Oktober 2026

Artificial Analysis says Mistral Large 4 is the most intelligent model from outside the US and China, achieving results comparable to DeepSeek V4.1 Flash (max) (Artificial Analysis)techmeme.com
Reflection's Beam becomes the most capable open-weight model built outside Chinathe-decoder.com
CATL and Tencent back Deepseek's ballooning funding round as the AI startup eyes a 2027 IPOthe-decoder.comIM MAGAZIN
DeepSeek to raise at least $12 billion in Tencent-backed fundingeconomictimes.indiatimes.com
Sources: DeepSeek is close to raising $12B+ in a round that could reach ~$14.9B, with Tencent and CATL as the biggest contributors, ahead of an early-2027 IPO (Bloomberg)techmeme.com
Nvidia-backed Reflection unveils first AI model to take on Chinese open modelseconomictimes.indiatimes.com

05. Oktober 2026

The 552B DeepSeek V4.1-Flash Model Offers A Peak Output Of 494 Tokens/Second When Powered By An At-Home Rig Spanning 4x NVIDIA DGX Spark Unitswccftech.com

01. Oktober 2026

DeepSeek and Huawei release open-source Ascend AI programming tools to reduce reliance on Nvidia ecosystemtomshardware.com
Firm rents four Nvidia H200s to test '80x cheaper' DeepSeek claimtomshardware.com

30. September 2026

DeepSeek, Huawei Build AI Chip Software to Challenge NVIDIAtechstrong.ai
DeepSeek and Huawei Target a Key Source of Nvidia’s A.I. Dominancenytimes.com
China's AI industry closes ranks as Deepseek ships open-source software for Huawei's Ascend chipsthe-decoder.com
DeepSeek, Huawei Build 128-Chip AI System to Challenge Nvidiaanalyticsinsight.net
China’s DeepSeek open-sources tools to help Huawei chips supplant Nvidia in AI | South China Morning Postscmp.com
DeepSeek says it has partnered with Huawei to develop programming tools for Huawei's Ascend chips, including TileLang, an open-source CUDA alternative (Reuters)techmeme.com
DeepSeek partners with Huawei to develop chip programming tools, reducing reliance on Nvidiaeconomictimes.indiatimes.com

29. September 2026

Alibaba, DeepSeek AI Agents Show Signs of Deception in New Testsanalyticsinsight.net

28. September 2026

Astra is terrible - anyone had bad reviews?community.openai.com

24. September 2026

Sources: DeepSeek's annualized revenue run rate hits $1B, up from less than $500M a few months ago, as it aims to finalize a ~$7.5B fundraise by late October (The Information)techmeme.com

23. September 2026

DeepSeek details DSec sandbox infrastructure for agent trainingtechnode.com

22. September 2026

New ClosedQuorum Windows malware uses AI for attack decisionsbleepingcomputer.com
Sources: DeepSeek and Moonshot are among the companies briefing the UN Security Council on AI risks and international security during the UN General Assembly (Laurie Chen/Reuters)techmeme.com
Sources: the Cyberspace Administration of China is investigating DeepSeek and Moonshot over alleged data leaks to Anthropic via Claude and has questioned staff (The Information)techmeme.com

21. September 2026

Sources: DeepSeek CEO Liang Wenfeng says training on Huawei chips is one of DeepSeek's biggest bets and Huawei is set to deliver training chips in Q4 or Q1 2027 (The Information)techmeme.com

18. September 2026

Engrams Embedding Entendre: Codesign for Efficient DRAM/SSD Offloadingnewsletter.semianalysis.com

17. September 2026

AI Race Explained: How China closed the gap on American artificial intelligence models; here's a detailed comparison between themetnownews.com

16. September 2026

DeepSeek Engineer Stuck In Self Destructive Loop – Says Building AI To Replace Him Is The Only Way To Stop OpenAI & Anthropic From Creating A “Cyberpunk 2077” Societywccftech.com

15. September 2026

DeepSeek AI engineer slams Anthropic, OpenAI over ‘pacing’ calls, invokes Nazi Germanyscmp.com
9 Things to Know About DeepSeek’s New 552B AI Modelanalyticsinsight.net
Anthropic Getting To AGI Is Like Hitler Getting The Atomic Bomb Before The Allies, Says DeepSeek Engineer Shengyu Liu In Viral Blogpostofficechai.com
Vera Rubin NVL72 inference tests show up to 7x better token throughput per MW vs. Blackwell on a 1.6T DeepSeek model, above Huang's 3x claim for 1T-3T LLMs (Bryan Shan/SemiAnalysis)techmeme.com

14. September 2026

Sooooo Slowwwwwwwww?!?!? What is going oncommunity.openai.com
China's DeepSeek hires dealmaker as CFO ahead of possible IPOeconomictimes.indiatimes.com
Accelerating Dropless MoE Training in JAX with NVIDIA Transformer Enginedeveloper.nvidia.com
DeepSeek begins limited test of AI voice interaction with four voice profilestechnode.com

11. September 2026

Anthropic Says Seven China-Based AI Labs Ran Industrial-Scale Claude Distillation Attacksthehackernews.com
How hackers used Claude for missiles, drone swarms, and surveillance, while Chinese labs mined it for training datathe-decoder.com
A cheap support model with an expensive agent's full context?community.openai.com
DeepSeek's new model sets a template for powerful LLMs that run leantheregister.com

10. September 2026

DeepSeek releases V4.1-Flash, says it outperforms flagship V4-Prosiliconangle.com
Moonshot, DeepSeek secretly routed user requests to Claude, Anthropic claimsscmp.com
Anthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeektechcrunch.com
DeepSeek Unveils V4.1-Flash Model with Architectural Upgrades, Price Cuts Ahead of Shanghai IPOtechstrong.ai
Anthropic details distillation efforts by Chinese companies, like Moonshot and DeepSeek, sending user queries to Claude via "transfer stations" outside China (Wall Street Journal)techmeme.com
DeepSeek Plans Shanghai IPO with CITIC Securities Backingaitechtrend.com
New Deepseek model V4.1-Flash cuts memory needs for AI agentsthe-decoder.com
DeepSeek debuts DeepSeek-V4.1-Flash, its smallest model built on a new Causal Encoder-Decoder architecture, with 552B backbone parameters and 1M-token context (Reuters)techmeme.com
DeepSeek releases Harness 0.1.5 with V4.1 Flash support, file uploads and sidebar previewstechnode.com
DeepSeek V4.1 Flash Matches GPT 5.6 Sol, Claude Opus 5 On Some Benchmarks At Substantially Lower Pricingofficechai.com
DeepSeek formally launches V4.1 Flash, routes V4 Pro requests to Flashtechnode.com
Six AI Teams, One Task: CodeFlowMu/FCoP Delivery and Failure Evidencecommunity.openai.com

09. September 2026

FBI, NSA warn Chinese AI companies like DeepSeek and Alibaba are reportedly carrying out 'industrial-scale' distillation campaigns to boost their modelstechradar.com
Databricks unveils adaptive AI retrieval model to cut search costs and latencyinfoworld.com
DeepSeek Harness Flaw Let AI Agents Disable Their Own File Sandbox Without Approvalthehackernews.com
DeepSeek reportedly hires CITIC Securities for Shanghai STAR Market IPOtechnode.com
China's DeepSeek taps CITIC Securities for domestic IPOeconomictimes.indiatimes.com
DeepSeek begins limited-time beta of V4.1 Flash multimodal modeltechnode.com
US claims Chinese AI companies’ core AI strategy is distilling American modelstheregister.com
The NSA, CISA, and FBI issue a joint advisory warning that Chinese AI companies, including DeepSeek, are conducting "industrial-scale" distillation campaigns (Reuters)techmeme.com

08. September 2026

DeepSeek is hiring 150 engineers, and none of them will touch a modelthenewstack.io
DeepSeek says it is seeking ~150 senior engineers in an "unprecedented" hiring spree to overhaul its backend systems, strained by high user demand and AI agents (Minxiao Chang/South China Morning Post)techmeme.com
DeepSeek embarks on ‘unprecedented’ hiring spree as it overhauls backend systemsscmp.com

07. September 2026

Well well well 5 hours limitcommunity.openai.com
DeepSeek reportedly plans 160,000-chip Huawei cluster in Inner Mongoliatechnode.com
Thailand pauses all datacenter builds and approvalstheregister.com

04. September 2026

DeepSeek plans big Huawei AI chip order to power new data centreeconomictimes.indiatimes.com
NVIDIA Is Reportedly Giving Up On China As A Lost Cause Amid An Unassailable Lead From Huawei, As DeepSeek Snubs H20 GPUs For 160,000 Units Of Huawei’s Ascend 950DT Chipswccftech.com
Deepseek plans the largest known Huawei chip cluster with 160,000 processors in Inner Mongoliathe-decoder.com
Sources: DeepSeek plans to use 160K+ of Huawei's Ascend 950DT chips at an Inner Mongolia data center, which would create one of the largest Huawei chip clusters (Mackenzie Hawkins/Bloomberg)techmeme.com

02. September 2026

Why I still choose GPT after trying Claude, Gemini and DeepSeekcommunity.openai.com

31. August 2026

DeepSeek’s first vision model vs. Gemini 3.7 Flash: It comes down to spend vs. speedthenewstack.io

27. August 2026

Sources: DeepSeek is set to soon raise $7.4B at a $74B valuation from existing and new investors to fund R&D and its expansion into computing infrastructure (Wall Street Journal)techmeme.com

26. August 2026

Claude Desktop can now easily run Qwen, DeepSeek and Kimi models — after Ollama’s first effort stalledthenewstack.io
Alibaba releases Qwen3.8-Flash-Next, targeting "ultimate cost efficiency"the-decoder.com
What are ‘open’ AI models?computerworld.com
Sources: DeepSeek generated $70.7M in revenue and posted a $106M net loss in the first seven months of 2026, ~10x its full-year 2025 revenue on a $139M net loss (The Information)techmeme.com

25. August 2026

Taiwanese cybersecurity firm warns that AI tools have more than doubled Chinese state-backed cyberattacksthe-decoder.com
Researchers detail the growing use of AI in cyberattacks across many Chinese state-linked groups, primarily using open-weight models like Kimi K3 and DeepSeek (Mark Anderson/Bloomberg)techmeme.com

24. August 2026

NVIDIA Vera Rubin Records A Massive 30x Increase In Throughput Per Watt Than Blackwell While Offering a 35x Token Cost Reduction Across Agentic AI Workloadswccftech.comIM MAGAZIN

23. August 2026

China’s overlooked internet army is quietly embedding AI into everyday businessesscmp.com
Sources: Nvidia plans to use its $6B deal with Poolside to build an open-weight AI model to compete with Chinese models like DeepSeek and Kimi (Robbie Whelan/Wall Street Journal)techmeme.com

21. August 2026

$16,000 Quad AI Geekom mini PC cluster gets DeepSeek V4 Flash treatment with 512GB RAM — reaches 14.61 tokens per secondtechradar.com
DeepSeek debuts multimodal language model competitive with Opus 4.8siliconangle.com
Deepseek releases experimental Flash vision model that rivals Opus 4.8 on agent benchmarksthe-decoder.com
DeepSeek Unveils Vision Model to Challenge Anthropic, Pitching Low Cost Against Top Performancetechstrong.ai
DeepSeek unveils an experimental version of its V4 Flash model that can understand visual prompts, saying it nears the performance of Anthropic's Opus 4.8 (Bloomberg)techmeme.com
DeepSeek Releases V4-Flash-Vision-Exp, Matches Opus 4.8 On Some Multimodal Benchmarksofficechai.com

19. August 2026

Snowflake Adds Ability to Dynamically Route Prompts Across Multiple AI Modelstechstrong.ai

18. August 2026

Accelerating AI innovation through open weightsinfoworld.com
Codex quota changes frustrate paying userscommunity.openai.com
Alibaba’s lightweight Qwen model takes on larger AI systems from OpenAI, DeepSeek, Zhipuscmp.com
At Beijing AI-themed bar, DeepSeek tokens come with the pintseconomictimes.indiatimes.com
DeepSeek V4 Pro 0813 vs GPT-5.6 Sol on DeepSWE: Cost, Coding, and Routingtogether.ai

17. August 2026

DeepSeek V4 Pro 0813 vs Claude Fable 5 on DeepSWE: Cost, Coding, and Routingtogether.ai

16. August 2026

Home province of DeepSeek, Moonshot founders seeks to retain, attract future AI talentscmp.com

15. August 2026

I Tested 6 Best Large Language Models in 2026aitechtrend.com

14. August 2026

DeepSeek's innovative harness treats everything as a plug-intheregister.com
How to effectively serve MTP-based Gemma 4 models for inference performancediscuss.google.dev
Understanding the economics of AI factoriesinfoworld.com
Launch of DeepSeek’s Harness marks its strategic pivot towards autonomous agentic AIscmp.com
DeepSeek releases DeepSeek Harness under the MIT license in developer preview, touting a design where "every capability is a plugin" that can be swapped out (Carl Franzen/VentureBeat)techmeme.com
Google cuts Gemini 3.7 Flash prices as enterprise AI economics diverge and Pro cadence slowsinfoworld.com
DeepSeek V4 Pro launches as US-China open AI model race intensifiestechwireasia.com
Against Claude Cowork, DeepSeek opens its open-source Harness to developerstechnode.com
DeepSeek to introduce peak and off-peak pricing for its APItechnode.com
How telcos can capture value in the AI economycomputerweekly.com
DeepSeek raises some V4 prices by more than 10x as AI demand strains capacitycomputerworld.com
DeepSeek raises some V4 prices by more than 10x as AI demand strains capacityinfoworld.com

13. August 2026

DeepSeek open sources an agent harness where everything is a pluginthenewstack.io
Deepseek ships improved V4 Pro, open-sources its agent software, and raises API pricesthe-decoder.com
DeepSeek raises prices, adding dynamic pricing, ahead of a potential IPO; V4-Flash output tokens go from $0.28/1M to $1.32 during peak hours and $0.66 off-peak (Saritha Rai/Bloomberg)techmeme.com
DeepSeek releases official V4 Pro model as it steps up expansioneconomictimes.indiatimes.com
DeepSeek raises API pricing for its V4 modelseconomictimes.indiatimes.com
DeepSeek launches V4-Pro, its most advanced model that rivals Kimi K3 on some benchmarks at much lower prices, costing $0.44/1M input and $0.87/1M output tokens (Juro Osawa/The Information)techmeme.com
DeepSeek v4-0813-Pro Benchmarks: Model Scores One Point Higher Than Flash On Artificial Analysis Intelligence Indexofficechai.com
DeepSeek’s updated V4 Pro AI model struggles on benchmarks, shines in cybersecurityscmp.com
DeepSeek V4 Pro API Update Adds Responses API Supporttechnode.com

12. August 2026

Microsoft's new MAI Code 1.1 Flash gets crushed by Deepseek on both price and performancethe-decoder.com
DeepSeek Expands Hiring for AI Data-Center Infrastructuretechnode.com

11. August 2026

Is DeepSeek safe?witness.ai

10. August 2026

V4-Flash vs. V4-Pro: DeepSeek promised better and cheaper. It’s true, but not how I expected.thenewstack.io

07. August 2026

DeepSeek Takes RMB141 Million Strategic Placement in Unitree IPOtechnode.com

06. August 2026

Claude Code is the fastest agent framework but costs nearly three times more than the cheapest rivalthe-decoder.com
Sources: DeepSeek has resumed its funding round, seeking $8B at a $74B valuation, after pausing talks following the leak of Liang Wenfeng's remarks to investors (Bloomberg)techmeme.com
Filing: DeepSeek has invested ~$20.8M in Unitree Robotics' Shanghai IPO and agreed to jointly develop AI models for humanoid machines (Eduardo Baptista/Reuters)techmeme.com
DeepSeek invests $20.8 million in Unitree's Shanghai IPOeconomictimes.indiatimes.com
DeepSeek says it plans to implement substantial price increases across its services; it currently charges $0.14/1M input and $0.28/1M output tokens for V4 Flash (Bloomberg)techmeme.com
DeepSeek Plans Significant API Price Increasestechnode.com
DeepSeek-V4 Flash 0731 vs GPT-5.6 Luna on DeepSWE: Cost and Codingtogether.ai

05. August 2026

Alibaba, DeepSeek push China’s AI model race towards lower costsartificialintelligence-news.com
DeepSeek reportedly reopens talks on a RMB50 billion second funding roundtechnode.com
DeepSeek V4 Flash tops OpenRouter weekly ranking with 7.22 trillion tokenstechnode.com
DeepSeek V4 Flash tops global weekly usage rankings with 7.22 trillion tokenstechnode.com

04. August 2026

Time to rethink US frontier AI model dominancecomputerweekly.com
DeepSeek-V4-Flash API launches on China’s National Supercomputing Internettechnode.com

03. August 2026

China turns up the heat with open model blitz as US model makers panictheregister.com
DeepSeek’s smaller model just outperformed its own flagshipthenewstack.io
China’s DeepSeek beefs up agentic AI with ‘harness’ tests as V4 model jolts Silicon Valleyscmp.com
Artificial Analysis: DeepSeek's V4-Flash costs $0.14/1M input and $0.28/1M output tokens, or $0.03 per test, far below Kimi K3's $0.86 and GPT-5.6 Sol's $1.86 (Eduardo Baptista/Reuters)techmeme.com
DeepSeek's new AI model is by far the cheapest of well-known models to run, research firm sayseconomictimes.indiatimes.com
Meta straps on a ‘Kick Me’ sign by mistakenly taking down video by India’s prime ministertheregister.com

02. August 2026

DeepSeek V4 Flash Outperforms Fable 5 On Terminal Bench While Being 99% Cheaperofficechai.com

31. Juli 2026

Hacker uses DeepSeek AI to autonomously attack vulnerable serversbleepingcomputer.com
New Deepseek Flash model matches OpenAI's GPT-5.6 Luna at roughly 60 percent lower costthe-decoder.com
DeepSeek’s V4 Flash Guts OpenAI’s Price War Within Hours, Matching Opus-Grade Output At $0.14 Per Million Tokens, As Moonshot Scales With 20,000 New NVIDIA GPUswccftech.com
DeepSeek V4 Flash scores 50 on the Artificial Analysis Intelligence Index, matching Gemini 3.6 Flash and up 10 points from the preview launch in April (Artificial Analysis)techmeme.com
Chinese Hacker Commands DeepSeek via Telegram to Launch Autonomous Attacksthehackernews.com
DeepSeek rolls out the official V4 Flash API in public beta, touting enhanced agent capabilities and benchmark scores "far surpassing" V4 Pro Preview (Newley Purnell/Bloomberg)techmeme.com
DeepSeek-v4-Flash-0731 Scores 50 On Artificial Analysis Intelligence Index, Creates Big Spike On Pareto Frontierofficechai.com
DeepSeek Releases DeepSeek-V4-Flash-0731, Gives Opus 4.8-Level Performance At A Fraction Of The Priceofficechai.com
DeepSeek puts V4-Flash API into public betatechnode.com

30. Juli 2026

Sources: DeepSeek plans to build a 1 GW data center in Inner Mongolia and aims to bring at least part of its capacity online by the end of 2027 or early 2028 (Bloomberg)techmeme.com

29. Juli 2026

Moonshot AI closes $3.5 billion funding round to hit $35 billion valuationeconomictimes.indiatimes.com

28. Juli 2026

DeepSeek V4 Flash, up to 32 tok/s on AMD Ryzen AI MAX+ 395lucebox.com

27. Juli 2026

DeepSeek reportedly halts funding round after Wenfeng’s commentssiliconrepublic.com
Our Compute Cost Was Lesser Than DeepSeek, Built 3 Models For $20 Million: India’s Sarvam AIofficechai.com

26. Juli 2026

Low profile, high AI ambition: What leaked comments reveal about DeepSeek’s Liang Wenfengscmp.com

25. Juli 2026

DeepSeek’s Normally Boring CEO Is Reportedly Spiraling After Supposed Leakgizmodo.com
Sources: DeepSeek told investors it is suspending its second funding round after remarks attributed to Liang Wenfeng on US-China AI competition went viral (Pei Li/Bloomberg)techmeme.comIM MAGAZIN
DeepSeek tells prospective investors of funding pause, Bloomberg News reportseconomictimes.indiatimes.comIM MAGAZIN
Ask HN: How to use agents via API cheaply?news.ycombinator.com
Investment with Chinese characteristics: how Beijing’s money is reshaping tech venturesscmp.com

24. Juli 2026

Top AIs invent same fake PyPl and npm package namesinfoworld.com

23. Juli 2026

DeepSeek CEO Believes NVIDIA Is Now “Digging Its Own Grave” Even As 1 NVIDIA GB300 GPU Equals 4 Huawei Acend 950 GPUswccftech.com
DeepSeek puts AGI research ahead of products and commercial growthtechnode.com

22. Juli 2026

FAQ Claude Mythos: Fähigkeiten, Zugang, Wettbewerber, Auswirkungencomputerwoche.de

21. Juli 2026

China’s trillion-parameter AI models may help enterprises with some applicationsinfoworld.com
Another ‘DeepSeek moment’? What China’s Kimi K3 means for the global AI industryscmp.com

17. Juli 2026

Moonshot AI unveils world’s largest open-source AI model as China narrows gap with US rivalsscmp.com

15. Juli 2026

DeepSeek founder Liang Wenfeng’s fortune rises to $36 billiontechnode.com
DeepSeek may file mainland IPO application this yeartechnode.com

14. Juli 2026

DeepSeek’s Liang Wenfeng tops Dario Amodei and Greg Brockman as richest AI foundereconomictimes.indiatimes.com
QuestMobile: China’s AI-native apps reach 499 million monthly active userstechnode.com

12. Juli 2026

China's DeepSeek developing its own AI chip, sources sayreuters.com

11. Juli 2026

DeepSeek-AI/DeepSeek-v3.2huggingbay.xyz

10. Juli 2026

Head to Head: Muse Spark 1.1 vs. DeepSeek-V4-Proruntimewire.com
China's 'Silicon Valley' in Hangzhou: Home of DeepSeek, Alibaba and Ant Group [video]youtube.com
Facing US export controls, China’s DeepSeek plans to make its own chipsarstechnica.com

09. Juli 2026

DeepSeek aims to make its own AI chipproactiveinvestors.com

08. Juli 2026

China Weighs Reversing Its Own NVIDIA Ban, May Let Alibaba, ByteDance & Others Buy 200,000 H200 Chipswccftech.comIM MAGAZIN
DeepSeek DSpark: The Speculative Decoding Trick Behind 400% Faster LLManalyticsvidhya.com
Anthropic's Claude Fable 5 dominates new industry benchmarks at a steep premiumthe-decoder.com
China plans to let top AI firms buy limited Nvidia H200 chips, the Information reportseconomictimes.indiatimes.com
Intel’s Arc Pro B70 Beats NVIDIA’s RTX 5090D In DeepSeek R1 AI LLM, Despite Costing A Quarter As Much, Offers Over 2000 Tokens/swccftech.com
DeepSeek Is Reportedly Building Its Own Inference Chip to Break Free From Both NVIDIA and Huaweiwccftech.com
DeepSeek begins in-house AI chip development to cut reliance on NVIDIA, sources saytechnode.com
DeepSeek accidentally built a working ransomware strain, experts note, 'What we are witnessing is a fundamental shift in how novel cyber attacks are born'techradar.com

07. Juli 2026

Report: China’s DeepSeek follows OpenAI in developing its own custom inference chipssiliconangle.com
U.S. Tech Companies Turn to Chinese AI Models as Domestic Costs Skyrockettechstrong.aiIM MAGAZIN
How did we make DeepSeek outperform Opustwitter.com
Facing US export controls, China's DeepSeek plans to make its own chipsarstechnica.com
A verification loop 4x'd DeepSeek's intelligence, matching Opus at 1/7 the costironbee.medium.com
Deepseek is designing its own AI chipthe-decoder.com
China's DeepSeek developing its own AI chipeconomictimes.indiatimes.com
Sources: DeepSeek is in the early stages of developing its own AI chip designed for inference, in a bid to reduce its reliance on Nvidia and Huawei hardware (Reuters)techmeme.com
Show HN: An always-fresh memory that learns your repo, so agents stop re-readinggithub.com

06. Juli 2026

DeepSeek V4 Is Earning Agentic Token Shareopenrouter.ai

05. Juli 2026

LongCat-2.0news.ycombinator.com

04. Juli 2026

DeepSeek drops another breakthrough [video]youtube.com

03. Juli 2026

Show HN: How clanker are you? A reverse Turing testhowclankerareyou.com

02. Juli 2026

DeepSeek reverses course with surcharge on peak-hour API usescmp.com

01. Juli 2026

Somebody told DeepSeek to build in-browser ransomware and it gleefully compliedtheregister.com
AI-Generated Browser Ransomware Abuses Chromium API on Windows and Androidthehackernews.com

30. Juni 2026

NVIDIA Slashes DeepSeek v4 Token Costs By Up To 5x Just One Month After Launch, Through Pure Blackwell Software Tuningwccftech.com
Why the AI future won’t be decided by algorithms and chatbotsscmp.com
Deepseek's DSpark boosts AI speed by up to 85 percent, a strategic win under tightening US export controlsthe-decoder.com
China debuts biggest AI model trained on local chips, as Meituan releases LongCat-2.0scmp.com
Show HN: No ads and noise from any page, get a clean AI reformat in one clickcode.intellios.ai
Fastllm: A LLM inference library that runs DeepSeek-V4 with 10GB VRAMgithub.com
DeepSeek to launch V4 in mid-July with new peak-time API pricingtechnode.com

29. Juni 2026

DeepSeek Releases Open-Source Inference Framework to Slash Compute Coststechstrong.ai
Show HN: Cline subscription plan to access GLM-5.2 at 2-5x discountcline.bot
Low-cost Chinese AI models like DeepSeek gain traction in the U.S.restofworld.org
DeepSeek V4 Peak Valley Pricing Changekucoin.com
DeepSeek details DSpark, a speculative decoding framework for its V4 models, saying it speeds up AI inference by up to 85% and was tested on Gemma and Qwen (Ben Jiang/South China Morning Post)techmeme.com

28. Juni 2026

Sina's open model VibeThinker-3B aims to show reasoning compresses well but factual knowledge doesn'tthe-decoder.com

26. Juni 2026

Show HN: Smart model routing directly in Claude, Codex and Cursorgithub.com
DeepSeek Is Recruitingapp.mokahr.com
China’s Zhipu AI sparks new ‘DeepSeek moment’ with cost-effective coding modelscmp.com

25. Juni 2026

Show HN: DeepSeek Flash inverted the economics of agent productsrtrvr.ai
We got DeepSeek-V4-Pro serving in 20 secondsinferize.ai
DeepSeek hiring spree: Chinese AI firm seeks newcomers as it pursues AGIscmp.com
Anthropic accuses Alibaba of using 25,000 fake accounts to scrape Claude AIcomputerworld.com
Anthropic accuses Alibaba of using 25,000 fake accounts to scrape Claude AIinfoworld.com

17. Juni 2026

Malicious JetBrains Plugins Steal AI API Keys as Chrome Extensions Capture Chatbot Chatsthehackernews.com
JetBrains Marketplace Supply Chain Attack: 15 Malicious AI Plugins & API Key Exfiltrationsatyamrastogi.com

08. Juni 2026

DeepSeek V4 forces rivals to slash prices, rattling China’s cloud providersscmp.com

05. Juni 2026

Huawei chips refine DeepSeek model in major leap for China’s AI self-reliancescmp.com

04. Juni 2026

DeepSeek in talks to raise $7 billion from Tencent, CATL and other investorstechnode.com

24. Mai 2026

DeepSeek V4 Pro tops global bang-for-buck ranking after 75% price cutscmp.com

11. Mai 2026

Serving DeepSeek-V4: why million-token context is an inference systems problemtogether.ai

04. Mai 2026

Deepseek V4 explained, and why it matters to your walletblog.lmcache.ai
Deepseek V4 explained, and why it matters to your walletblog.lmcache.ai

03. Mai 2026

DeepSeek’s Sequelnytimes.com
Caisi Evaluation of DeepSeek V4 Pronist.gov
China is falling behind in the AI race, according to a US government benchmarkthe-decoder.com
Xiaomi's open-weight MiMo-V2.5-Pro takes aim at Claude Opus with hours-long autonomous codingthe-decoder.com
An evaluation by NIST's CAISI says DeepSeek V4 Pro lags behind leading US AI models by about eight months and is the most capable Chinese AI model to date (NIST)techmeme.com

02. Mai 2026

AgInTiFlow, a local web and CLI agent workspace using low-cost DeepSeeknpmjs.com
Show HN: Filling PDF forms with AI using client-side tool callingcopilot.simplepdf.com

01. Mai 2026

DeepSeek v4, and the end of the OpenAI/Microsoft AGI clausesimonw.substack.com
DeepSeek V4–almost on the frontier, a fraction of the pricesimonwillison.net
DeepSeek V4 Flash and V4 Pro in Microsoft Foundrytechcommunity.microsoft.com

30. April 2026

Italy closes antitrust probes into AI firms after commitments on 'hallucination' riskseconomictimes.indiatimes.com
Italy closes antitrust probes into AI firms after commitments on 'hallucination' riskseconomictimes.indiatimes.com
DeepSeek begins limited testing of vision modetechnode.com
Allow Users to Customize Chat Window Width in ChatGPTcommunity.openai.com
DeepSeek continues to challenge closed-source competitors with DeepSeek-V4dataphoenix.info

29. April 2026

Ask HN: Are you OK with DeepSeek and other labs reading your data?news.ycombinator.com
‘The whale can now see’: DeepSeek adds AI vision in major movescmp.com
Big Chinese tech firms scramble to secure Huawei AI chips after DeepSeek V4 launcheconomictimes.indiatimes.com
A 3D Flappy Bird side-scroller game built with DeepSeek V4 Proannajc.com
DeepSeek-V4 Pro now available on Together AItogether.ai

28. April 2026

DeepSeek V4 Pro: Validating Frontier Models for Productionfireworks.ai
Tencent, Alibaba to back DeepSeek at $20B+ valuation: reporttechfundingnews.com
DeepSeek Unveils Newest Flagship AI Model a Year After Upending Silicon Valleybloomberg.com
DeepSeek V4 is a display of Huawei AI chip's capabilitiestechnologyreview.com
DeepSeek V4 report shows multiple R&D staff departurestechnode.com
Show HN: Lightport – open-source AI gatewaynpmjs.com

27. April 2026

China’s DeepSeek prices new V4 AI model at 97% below OpenAI’s GPT-5.5scmp.com
DeepSeek-V4 arrives with near SotA intelligence at 1/6th the costventurebeat.com
Show HN: Lightport – AI gateway that makes LLM providers OpenAI-compatiblegithub.com
AI did what? ChatGPT's big upgrade, Google’s plans to make AI invisible, DeepSeek’s return and more of the week’s most surprising developmentstechradar.com
The Download: DeepSeek's latest AI breakthrough, and the race to build world motechnologyreview.com
First DeepSeek V4 Flash-Base-Int4 Quanthuggingface.co
‘Extraction and distillation’: US State Department upgrades AI theft accusations to target China’s Deepseek, Moonshot AI and MiniMaxtechradar.com
DeepSeek launches V4 model adapted for Huawei AI chipstechwireasia.com
DeepSeek takes step away from Nvidia amid export curbs on Chinascmp.com
Show HN: Another experiment with an Erdos problem and LLMsnews.ycombinator.com
DeepSeek V4 becomes default model for OpenClawtechnode.com

26. April 2026

DeepSeek drops input cache price to 1/10thxcancel.com
DeepSeek's new models are so efficient they'll run on a toaster by which we meantheregister.com
The disappearing AI middle classthenewstack.io
DeepSeek V4 plays Go on a 9x9 boardchat.deepseek.com
DeepSeek launches 1.6 trillion parameter V4 on Huawei chips as U.S. escalates AI theft accusations — U.S. gov't alleges IP theft by DeepSeek and other Chinese AI firmstomshardware.com
NVIDIA Beats Everyone To DeepSeek V4 With Day-0 Blackwell Support, Pushing 3,500 Tokens Per Second On 1.6T Modelswccftech.com

25. April 2026

DeepSeek-V4 on Day 0: From Fast Inference to Verified RL with SGLang and Mileslmsys.org
US State Dept orders global warning about alleged AI thefts by DeepSeek, other Chinese firmseconomictimes.indiatimes.com

24. April 2026

Build with DeepSeek V4 Using NVIDIA Blackwell and GPU-Accelerated Endpointsdeveloper.nvidia.com
I audited Google AI Studio’s frontend. Here’s what I founddiscuss.google.dev
The Coding Assistant Breakdown: More Tokens Pleasenewsletter.semianalysis.com
DeepSeek's new models are so efficient they'll run on a toaster ... by which we mean Huawei's NPUstheregister.com
DeepSeek's new models are so efficient they'll run on a toaster ... by which we mean Huawei's NPUsgo.theregister.com
DeepSeek open-sources V4 large language model seriessiliconangle.com
AI Arms Race Accelerates With New Models from OpenAI, DeepSeek and Anthropiccnet.com
DeepSeek V4 Squeezes Million-Token Context Into 10% of V3.2’s Memory, Escalating China’s AI Efficiency War With OpenAIwccftech.com
DeepSeek V4 Pro Becomes Second-Highest Rated Open Model On Artificial Analysis Index With Score Of 52officechai.com
Former OpenAI research scientist launches new AI model for Tencentcomputerworld.com
Former OpenAI research scientist launches new AI model for Tencentinfoworld.com
DeepSeek’s New V4 Model Signals Shift Toward Hardware Independencetechstrong.ai
DeepSeek Says It Is 3-6 Months Behind US Frontier Labsofficechai.com
China’s DeepSeek releases new AI model V4. Here’s everything to know as the AI race speeds upeuronews.com
DeepSeek-V4, the Chinese AI model adapted for Huawei chipseconomictimes.indiatimes.com
DeepSeek previews new AI model that ‘closes the gap’ with frontier modelstechcrunch.com
DeepSeek-V4: The Most Powerful Open-Source Model Everanalyticsvidhya.com
Investors rotate to China’s chipmakers as DeepSeek intensifies AI competitionscmp.com
China’s DeepSeek previews new AI model a year after jolting US rivalstheverge.com
After eyeing $20B in valuation, DeepSeek launches its long-awaited model: What we know so fartechfundingnews.com
China’s DeepSeek Unveils V4 Preview, Eyes $20 Billion Valuation as Rivals Tumbleciolookmedia.com
China’s DeepSeek unveils long-awaited V4 AI modelsiliconrepublic.com
DeepSeek-V4 preview now available with open-source accesstechnode.com
DeepSeek is back with V4, slashing agentic AI coststechzine.eu
Huawei Ascend supernode to support Deepseek V4economictimes.indiatimes.com
DeepSeek Releases DeepSeek V4-Pro & V4-Flash, Delivers GPT 5.4 & Opus 4.6-Level Performance At Fraction Of The Priceofficechai.com
DeepSeek unveils next-gen AI model as Huawei vows ‘full support’ with new chipsscmp.com
DeepSeek-V4: a million-token context that agents can actually usehuggingface.co

23. April 2026

Alibaba, Tencent Eye Major Investment in DeepSeek: Reporttechstrong.ai

18. April 2026

DeepSeek Security Concerns Every Enterprise Should Understandwitness.ai

08. April 2026

Experimenting with GPUs: GKE managed DRANET and Inference Gateway AI Deploymentcloud.google.com
China’s DeepSeek adds expert chatbot mode ahead of much-awaited V4 releasescmp.com
DeepSeek V4 may launch this month, test interface suggests Vision and Expert modestechnode.com

06. April 2026

DeepSeek V4 points to growing use of Huawei chips in AI modelstechwireasia.com

03. April 2026

Deepseek v4 will reportedly run entirely on Huawei chips in a major win for China's AI independence pushthe-decoder.com

31. März 2026

Beetroot - Windows clipboard manager with built-in OpenAI text transformscommunity.openai.com

30. März 2026

Show HN: I made a cheaper alternative to Claude Code or Codex CLIsweetcli.com
Show HN: //Beforeyouship is a pre-build tool to estimate the LLM costllm-architecture-cost-modeler.vercel.app
DeepSeek Outage: Chinese AI chatbot suffered longest disruption since viral rise in early 2025; check updatesetnownews.com

27. März 2026

Reflection eyes $2.5B raise at $25B valuation from JPMorgan, Disruptive to counter DeepSeektechfundingnews.com
Five techniques to reach the efficient frontier of LLM inferencecloud.google.com

26. März 2026

China bans Manus founders from leaving country after Meta acquires AI startup and reviews dealeuronews.com
I tested Meta AI vs Deepseek with 5 real-world prompts — one completely dominatedtomsguide.com
OpenRouter data: lower-cost Chinese AI models by groups such as DeepSeek and MiniMax have surpassed US rivals in token consumption since last month (Zijing Wu/Financial Times)techmeme.com

25. März 2026

I ran 3,360 safety tests on GPT-4o, Claude, Grok, DeepSeek, Geminigithub.com

24. März 2026

China’s use of open‐source AI threatens the US lead in AI development, US Commission warnscomputerworld.com

19. März 2026

Multiverse Computing pushes its compressed AI models into the mainstreamtechcrunch.com

14. März 2026

I built an Agentic Writing Environment using DeepSeek to replace Word processorsminotauris.app

13. März 2026

Comparing GPT-5 Pro and earlier LLMs for developerscommunity.openai.com

12. März 2026

Show HN: A graph of story fragments shaped by reader votesslopism.art
Microsoft is pushing for more Africans to adopt its AI tools as it competes with DeepSeek: training 3M people, partnering with telecom MTN to sell 365, and more (Loni Prinsloo/Bloomberg)techmeme.com

11. März 2026

AI chatbots help plot attacks, study shows: ‘happy (and safe) shooting!’scmp.com
Nvidia Will Spend $26 Billion to Build Open-Weight AI Models, Filings Showwired.com

07. März 2026

Is AI Distillation Theft or Just How Knowledge Evolves?maxim-blog.ghost.io

05. März 2026

Malicious AI Assistant Extensions Harvest LLM Chat Historiesmicrosoft.com
Show HN: Nemilia – multi-agent AI workspace in a single HTML file, no back endgithub.com
Show HN: Nemilia – multi-agent AI workspace in a single HTML file, no back endgithub.com

04. März 2026

Show HN: ContextCache – Cache tool schema KV states, skip 99% of prefill tokensgithub.com
‘Two sessions’, two paths: China prioritises AI integration amid US tech rivalryscmp.com

03. März 2026

Inside the optimization of FP8 training on Ironwooddiscuss.google.dev

02. März 2026

Implementing GRPO with NVIDIA NeMo-RL on Google Kubernetes Enginediscuss.google.dev
China’s New Study Buddy: The A.I. Chatbotnytimes.com
DeepSeek plans V4 multimodal model release this week, sources saytechnode.com

01. März 2026

The Attention Arms Race: How Modern Open-Source LLMs Are Reinventing the Transformer's Coremaxim-blog.ghost.io
DeepSeek V4 Is About to Test America’s AI Lead: What We Know Before Launchumesh-malik.com

26. Februar 2026

DeepSeek withholds latest AI model from US chipmakers including Nvidia, sources sayeconomictimes.indiatimes.com

24. Februar 2026

DeepSeek trained AI model on Nvidia Blackwell chip despite US export ban,official saysdev.to
China's DeepSeek trained AI model on Nvidia's best chip despite US ban, official saysthehindu.com
Anthropic Says Chinese AI Firms Used 16 Million Claude Queries to Copy Modelthehackernews.com
OpenAI, Anthropic accuse Chinese rivals of mass AI data theftthehindu.com
Optimize ChatGPT for Long Conversations with Automatic Summarization and Adaptive Context Managementcommunity.openai.com
The $100M AI Heist: How DeepSeek Stole Claude's Brain With 16 Million Fraudulent API Callsumesh-malik.com

23. Februar 2026

US AI giant accuses Chinese rivals of mass data thefttheguardian.com
Anthropic accuses Chinese AI labs of mining Claude as US debates AI chip exportstechcrunch.com

21. Februar 2026

Here’s How NVIDIA’s Blackwell Ultra GB300 AI Racks Are Dominating Long-Context DeepSeek Workloads, Delivering Impressive Gains Versus GB200wccftech.com

17. Februar 2026

Alibaba Qwen 3.5: E-com giant unveils improved version - Know all about LLM model and latest updatesetnownews.com
Alibaba unveils new Qwen3.5 model for 'agentic AI era'thehindu.com

13. Februar 2026

OpenAI flags DeepSeek model copying concerns in AI racetechwireasia.com
DeepSeek-V3.2 on GB300: Performance Breakthroughblog.vllm.ai

12. Februar 2026

ChatGPT’s Translation Skills Parallel Most Human Translatorsspectrum.ieee.org

06. Februar 2026

How we cut Vertex AI latency by 35% with GKE Inference Gatewaycloud.google.com
China fines firms for fake ChatGPT and DeepSeek services amid tightening AI governancescmp.com
What do LLMs think when you don't tell them what to think about?together.ai

04. Februar 2026

The new AI lab that's coming for DeepSeeksources.news

30. Januar 2026

Nvidia says Chinese military dependence on American tech would be 'nonsensical,' following US govt agency's claims it assisted Deepseek with training AI models — says Admin's critics 'are unintentionally promoting the interests of foreign competitors'tomshardware.com

28. Januar 2026

DeepSeek releases OCR 2 with new visual encoding architecture, targeting more human-like machine visiontechnode.com

27. Januar 2026

Tech Lifebbc.co.uk
DeepSeek taps Alibaba open-source AI technology to boost OCR performancescmp.com

21. Januar 2026

DeepSeek Reportedly Prepares New Flagship AI Model Ahead of Lunar New Yeartechnode.com

19. Januar 2026

DeepSeek a year on: how a little-known Chinese start-up sparked a global AI arms racescmp.com

16. Januar 2026

Epoch AI 2025 impact reportepoch.ai

14. Januar 2026

DeepSeek proposes a workaround to train bigger AI models with less powerful chipstechwireasia.com

13. Januar 2026

Chinese Neuralink Rival BrainCo Files for Hong Kong IPO, Sources Saytechnode.com

11. Januar 2026

I tested Gemini 3 Flash vs. DeepSeek with 9 prompts — the winner surprised metomsguide.com

08. Januar 2026

DeepSeek to customise namesake chatbot for Italian users following probescmp.com

07. Januar 2026

8 plots that explain the state of open modelsinterconnects.ai

06. Januar 2026

China’s DeepSeek adds advanced ‘thinking’ feature to chatbot amid buzz over next modelscmp.com

05. Januar 2026

Latest open artifacts (#17): NVIDIA, Arcee, Minimax, DeepSeek, Z.ai and others close an eventful year on a high noteinterconnects.ai
Another Chinese quant fund joins DeepSeek in AI race with model rivalling GPT-5.1, Claudescmp.com

03. Januar 2026

Tech war: China takes confident strides to develop more AI innovation in 2026scmp.com

23. Dezember 2025

Top 10 Data Insights and Gradient Updates of 2025epoch.ai

22. Dezember 2025

Sharing a Robust Root Cause Analysis (RCA) Prompt – Systematic, Evidence-Based Troubleshootingcommunity.openai.com

05. Dezember 2025

What makes DeepSeek-V3.2 so efficient?bdtechtalks.com

01. Dezember 2025

DeepSeek Releases New Reasoning Models to Take On ChatGPT and Geminicnet.com
Together AI delivers fastest inference for the top open-source modelstogether.ai

28. November 2025

DeepSeek and Alibaba urge more transparency in AI governancetechwireasia.com

25. November 2025

Local LLMs are how nerds now justify a big computer they don't needworld.hey.com

24. November 2025

Qwen AI hits 10m+ downloads as Alibaba disrupts the AI marketartificialintelligence-news.com

27. Oktober 2025

On-Policy Distillationthinkingmachines.ai

21. Oktober 2025

DeepSeek releases new OCR model capable of generating 200,000 pages daily on a single GPUtechnode.com

10. Oktober 2025

AdapTive-LeArning Speculator System (ATLAS): A New Paradigm in LLM Inference via Runtime-Learning Acceleratorstogether.ai

02. Oktober 2025

U.S. Commerce Sec. Lutnick says American AI dominates DeepSeek, thanks Trump for AI Action Plan — OpenAI and Anthropic beat Chinese models across 19 different benchmarkstomshardware.com

01. Oktober 2025

Why AI phishing detection will define cybersecurity in 2026artificialintelligence-news.com

30. September 2025

DeepSeek Releases V3.2-Exp Experimental Model, Cuts API Prices by Over 50%technode.com

29. September 2025

DeepSeek releases ‘sparse attention’ model that cuts API costs in halftechcrunch.com

15. September 2025

MIT Technology Review releases 2025 ’50 Smartest Companies’ list, recognizes Deepseek, Game Science and Unitree Roboticstechnode.com

27. August 2025

DeepSeek-V3.1: Hybrid Thinking Model Now Available on Together AItogether.ai

26. August 2025

Tesla taps ByteDance, DeepSeek to power AI in Chinatechnode.com

21. August 2025

DeepSeek open-sources new V3.1 modeltechnode.com

15. August 2025

DeepSeek: The Chinese startup challenging Silicon Valleyartificialintelligence-news.com

14. August 2025

DeepSeek denies rumors of August launch for R2 AI modeltechnode.com

27. Juni 2025

German data protection official wants Apple, Google to remove DeepSeek from the country’s app storestechcrunch.com
DeepSeek AI model faces delays amid US chip export restrictionstechnode.com

13. Juni 2025

Alibaba Chairman Joe Tsai: DeepSeek’s rise jolted us into actiontechnode.com

16. Mai 2025

DeepSeek reveals cost-cutting methods for V3 large model training in new papertechnode.com

08. Mai 2025

DeepSeek credits Tencent for major performance boost in open-source framework DeepEPtechnode.com

16. April 2025

Alibaba-backed Zhipu AI preps IPO, first among China’s top six AI unicornstechnode.com

03. April 2025

Alibaba’s Qwen2.5-Omni tops Hugging Face’s open-source Ai model listtechnode.com

02. April 2025

DeepSeek files patent for smarter, more efficient AI data collectiontechnode.com

29. März 2025

DeepSeek’s popularity fuels concerns over misinformationtechnode.com

28. März 2025

DeepSeek founder Liang Wenfeng joins global billionaires listtechnode.com

26. März 2025

Tencent upgrades Yuanbao AI with self-developed model Hunyuan T1technode.com

25. März 2025

DeepSeek upgrades V3 model with more parameters, open-source shifttechnode.com

21. März 2025

01.AI founder Kai-Fu Lee predicts the endgame of China’s AI models, names DeepSeek as frontrunnertechnode.com

04. März 2025

Luchen Technology becomes first to drop DeepSeek API over cost concernstechnode.com
Tencent’s Yuanbao tops Apple’s China App Store, surpassing DeepSeektechnode.com

02. März 2025

ByteDance appoints new leader for AI video tool Jimeng, considers DeepSeek integrationtechnode.com

27. Februar 2025

Demand for NVIDIA H20 chips surges as Chinese companies adopt DeepSeek’s AI models: reporttechnode.com

26. Februar 2025

DeepSeek accelerates launch of next-gen AI model amid China’s AI pushtechnode.com

25. Februar 2025

DeepSeek reopens access to its API after three-week pausetechcrunch.com

24. Februar 2025

Tencent’s AI assistant Yuanbao rises to second on Apple’s free app chart, behind DeepSeektechnode.com
DeepSeek announces open-source initiative and revealed FlashMLA modeltechnode.com

21. Februar 2025

Weibo starts internal testing of DeepSeek R1 model for its intelligent search featuretechnode.com

19. Februar 2025

The AI app wars: a quick take on 2025technode.com

17. Februar 2025

Baidu Search integrates DeepSeek and Large Model ERNIE for advanced searchtechnode.com
WeChat integrates AI Search with DeepSeek, seeks to allay concerns over user privacytechnode.com

14. Februar 2025

DeepSeek founder Liang Wenfeng is reportedly set to meet with China’s Xi Jinpingtechcrunch.com
NetEase’s martial arts game Justice Mobile introduces DeepSeek-powered charactertechnode.com

13. Februar 2025

Vivo announces integration of DeepSeek into its mobile operating systemtechnode.com

12. Februar 2025

Torque Clustering: "Autonomous AI on the horizon"newatlas.com
General Motors, Mercedes-Benz among first global automakers to integrate DeepSeek AI modelstechnode.com
Apple tests DeepSeek model but shifts to Alibaba for AI features in Chinatechnode.com
Multiple third-party platforms integrate DeepSeek – has Chinese AI finally entered its era?technode.com

11. Februar 2025

Apple reportedly partners with Alibaba after rejecting DeepSeek for China AI launchtechcrunch.com

10. Februar 2025

DeepSeek-V3 ends promotional pricing, updates API service ratestechnode.com
Elon Musk comments on China’s DeepSeek at WELT summittechnode.com

08. Februar 2025

Alibaba denies it has made $1 billion investment in DeepSeektechnode.com
China’s Geely, Dongfeng to be first automakers to embrace DeepSeektechnode.com

07. Februar 2025

DeepSeek ban? China data transfer boosts security concernsartificialintelligence-news.com

06. Februar 2025

OpenAI now reveals more of its o3-mini model’s thought processtechcrunch.com
Moore Threads deploys DeepSeek distilled model for high-performance AI inference on domestic GPUstechnode.com
South Korea blocks DeepSeek on government devices, announces launch of “national AI committee”technode.com

03. Februar 2025

DeepSeek V3 Explained: Advancing AI Agents with Enterprise-Grade Reasoningallganize.ai

31. Januar 2025

DeepSeek Jailbreak’sadversa.ai

30. Januar 2025

E092 - DeepSeek for XR, Samsungs XR-Headset Moohan im Test, der Google-HTC-VIVE-Deal, und die Frage: Befinden wir uns gerade im XR-Herbst?metaverse-podcast.de
US investigates DeepSeek over national security concernstechnode.com
How to deploy and fine-tune DeepSeek models on AWShuggingface.co

29. Januar 2025

Microsoft brings a DeepSeek model to its cloudtechcrunch.com
DeepSeek limits registrations amid large-scale cyberattacktechnode.com
DeepSeek releases new models Janus-Pro and JanusFlow on Lunar New Year’s Evetechnode.com

28. Januar 2025

Hugging Face researchers are trying to build a more open version of DeepSeek’s AI ‘reasoning’ modeltechcrunch.com
Trillion-dollar disruptor: China's DeepSeek upends AI world overnightnewatlas.com

27. Januar 2025

DeepSeek claims its ‘reasoning’ model beats OpenAI’s o1 on certain benchmarkstechcrunch.com
Alibaba’s Qwen team releases AI models that can control PCs and phonestechcrunch.com
DeepSeek AI assistant surpasses ChatGPT on US App Storetechnode.com
DeepSeek ‘punctures’ AI leaders’ spending plans, and what analysts are sayingtechcrunch.com

26. Januar 2025

Silicon Valley stunned as China’s DeepSeek-R1 surpasses US AI giants, undercuts coststechnode.com

20. Januar 2025

DeepSeek-R1 reasoning models rival OpenAI in performanceartificialintelligence-news.com

27. Dezember 2024

DeepSeek launches new AI model with 671 billion parameters, rivaling GPT-4otechnode.com

14. Dezember 2024

‘Reasoning’ AI models have become a trend, for better or worsetechcrunch.com

20. November 2024

A Chinese lab has released a ‘reasoning’ AI model to rival OpenAI’s o1techcrunch.com
Modelldaten und Nachrichten aus BrunoSan AI-News, OpenRouter und Hugging Face. Benchmark-Werte stammen aus den jeweils verlinkten Quellen; Testbedingungen können abweichen. Modelle bleiben auch ohne Benchmark-Werte sichtbar.