Request a Pilot
‹ PublicAI Index

The LLM benchmark aggregator.

DeepSeek V4

DeepSeek

Strongest in Science (#30 of 122), weakest in Mathematics (#125 of 152). Above par in 4 of 11 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 5 and Claude Fable 5.1 and ahead of Kimi K2 and Claude Sonnet 4.

§ 1 · Profile

What it is good at

Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.

Knowledge53.1#60/197
Academic knowledge57.5#32/123
Coding49.8#88/179
Agentic coding49.8#86/173
Professional49.3#106/194
Finance51.1#90/162
Legal48.8#101/173
Medical47.2#102/149
Reasoning47.6#118/195
Science59.3#30/122
Mathematics38.3#125/152

§ 2 · Head to head

What it beats, and what beats it

The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for DeepSeek V4, left for the other.

Command R+won 7 · lost 0
Mistral Large 2.1won 8 · lost 0
Mistral Smallwon 8 · lost 0
Gemini 1.5 Flashwon 8 · lost 0
NVIDIA Nemotron 3.5 Lightning 30B A3Bwon 8 · lost 0
Mercury 2.5won 8 · lost 0
Jamba 1.6 Miniwon 9 · lost 0
Jamba 1.6 Largewon 9 · lost 0
Mistral Small 3.1 24B Instructwon 9 · lost 0
Llama 4 Scout Instructwon 9 · lost 0
Gemini 1.5 Prowon 9 · lost 0
GPT-4o miniwon 10 · lost 0
Laguna XS.2won 10 · lost 0
Laguna M.1won 10 · lost 0
MiMo V2.5won 10 · lost 0
Claude 3.5 Haikuwon 10 · lost 0
Mistral Medium 3.5won 11 · lost 0
GPT-4.1 nanowon 10 · lost 1
MiMo V2.5 Prowon 10 · lost 1
GPT-5 nanowon 8 · lost 1
Llama 3.3 70B Instructwon 7 · lost 1
Grok 2won 7 · lost 1
Llama 3.3 Nemotron Superwon 6 · lost 1
Gemma 4 31Bwon 6 · lost 1
Command Awon 9 · lost 2
Grok 4.20won 9 · lost 2
Claude Haiku 4.5won 9 · lost 2
Gemini 3.1 Flash-Litewon 9 · lost 2
MiniMax M2.7won 9 · lost 2
Llama 4 Maverick Instructwon 9 · lost 2
Gemini 3.5 Flash-Litewon 9 · lost 2
Ling 3.0 Flashwon 8 · lost 2
Gemini 2.5 Flash-Litewon 7 · lost 2
Mistral Medium 3.1won 6 · lost 2
Grok 4.1 Fastwon 8 · lost 3
NVIDIA Nemotron 3 Ultra 550B A55Bwon 8 · lost 3
GPT OSS 120Bwon 8 · lost 3
GPT-5.4 nanowon 8 · lost 3
Gemini 2.0 Flashwon 8 · lost 3
Inkling Smallwon 8 · lost 3
GLM-5.1won 8 · lost 3
DeepSeek V3.2won 8 · lost 3
GLM-4.7won 8 · lost 3
Qwen3.6 27Bwon 5 · lost 2
GPT-5.4 miniwon 7 · lost 3
Magistral Small 1.2won 6 · lost 3
Magistral Medium 1.2won 6 · lost 3
GPT OSS 20Bwon 6 · lost 3
Mistral Large 3won 7 · lost 4
GPT-4owon 7 · lost 4
MiniMax M2.5won 7 · lost 4
DeepSeek V3won 7 · lost 4
O3 Miniwon 7 · lost 4
Claude 3.5 Sonnetwon 6 · lost 4
DeepSeek R1won 6 · lost 4
Grok 4.3won 6 · lost 5
Grok 4 Fastwon 6 · lost 5
Kimi K2.5won 6 · lost 5
GLM-4.6won 6 · lost 5
Qwen3 Maxwon 6 · lost 5
Inklingwon 6 · lost 5
Qwen3.6 Pluswon 6 · lost 5
MiniMax M3won 6 · lost 5
GLM-5won 6 · lost 5
Qwen3.5 Flashwon 6 · lost 5
GPT-4.1 miniwon 6 · lost 5
MiniMax M2.1won 6 · lost 5
Claude Sonnet 4won 6 · lost 5
Kimi K2won 6 · lost 5
GPT-5 miniwon 5 · lost 6
Qwen 3.5 Pluswon 5 · lost 6
Kimi K2 Thinkingwon 5 · lost 6
GPT-5.1won 5 · lost 6
Gemini 2.5 Prowon 5 · lost 6
Claude Opus 4won 5 · lost 6
GLM-4.5won 4 · lost 5
Gemini 3 Flashwon 4 · lost 7
GLM-5.3 Flashwon 4 · lost 7
Grok 4won 4 · lost 7
GPT-4.1won 4 · lost 7
Grok 3won 4 · lost 7
Qwen3.8 27Bwon 4 · lost 7
Gemini 2.5 Flashwon 4 · lost 7
Claude 3.7 Sonnetwon 4 · lost 7
Qwen3 235B A22B Instructwon 4 · lost 7
O4 Miniwon 4 · lost 7
GPT-6 Lunawon 3 · lost 6
Step 5won 2 · lost 5
Claude Opus 4.5won 3 · lost 8
Claude Sonnet 4.5won 3 · lost 8
Kimi K2.6won 3 · lost 8
Claude Sonnet 4.6won 3 · lost 8
GPT-5won 3 · lost 8
Grok 3 Miniwon 3 · lost 8
GLM-5.2won 3 · lost 8
O1won 3 · lost 8
GLM-5.3won 3 · lost 8
O3won 3 · lost 8
Claude Opus 4.1won 2 · lost 7
Claude Haiku 5.5won 2 · lost 7
GPT-6 Solwon 2 · lost 7
Qwen3.7 Maxwon 2 · lost 9
Gemini 3.1 Prowon 2 · lost 9
Gemini 3 Prowon 2 · lost 9
Muse Sparkwon 2 · lost 9
Gemini 3.5 Flashwon 2 · lost 9
Gemini 3.6 Flashwon 2 · lost 9
GPT-5.2won 2 · lost 9
DeepSeek V4 Flashwon 2 · lost 9
Claude Sonnet 5won 2 · lost 9
DeepSeek V4 Prowon 2 · lost 9
Mistral Large 4won 1 · lost 7
GPT-6.1 Solwon 1 · lost 8
Claude Opus 4.6won 1 · lost 10
GPT-5.4won 1 · lost 10
GPT-5.6 Lunawon 1 · lost 10
GPT-5.6 Terrawon 1 · lost 10
Ember 1won 0 · lost 7
Muse Spark 1.3won 0 · lost 7
Muse Wonderwon 0 · lost 7
MiMo V2.6 Prowon 0 · lost 8
Grok 4.7won 0 · lost 8
MiMo V2.6 Flashwon 0 · lost 8
Hy4won 0 · lost 8
Muse Wonder Fullwon 0 · lost 8
Gemini 4 Argonwon 0 · lost 8
Claude Sonnet 5.5won 0 · lost 9
DeepSeek V4.1 Flashwon 0 · lost 9
GPT-6 Astrawon 0 · lost 9
Claude Opus 5.5won 0 · lost 9
Muse Spark 1.2won 0 · lost 10
Gemini 3.8 Flashwon 0 · lost 11
Gemini 3.7 Flashwon 0 · lost 11
Kimi K3won 0 · lost 11
Qwen3.8 Maxwon 0 · lost 11
Grok 4.5won 0 · lost 11
Claude Opus 4.7won 0 · lost 11
Muse Spark 1.1won 0 · lost 11
GPT-5.5won 0 · lost 11
Claude Opus 4.8won 0 · lost 11
Grok 4.6won 0 · lost 11
GPT-5.6 Solwon 0 · lost 11
Claude Fable 5won 0 · lost 11
Claude Fable 5.1won 0 · lost 11
Claude Opus 5won 0 · lost 11

§ 3 · Sources

Where the numbers come from

1 publication, 14 figures. Every one links to the page it was read from.

Vals.ai ↗14 measuresread 2026-11-10
Vals · Legal Research Bench 23.08%Vals · CaseLaw 59.38%Vals · LegalBench 80.32%Vals · Finance Agent 44.08%Vals · CorpFin 61.38%Vals · TaxEval 72.08%Vals · MedCode 40.45%Vals · MedScribe 75.14%Vals · SWE-bench Verified 77.4% (Mini-SWE-agent)Vals · Vibe Code Bench 49.93% (OpenHands)Vals · Code Migration 26.2%Vals · GPQA Diamond 89.39%Vals · MMLU Pro 87.25%Vals · ProofBench 16%

Badge

PublicAI Index badge for DeepSeek V4[![PublicAI Index](https://publicai.io/model-index/badge?model=deepseek-v4)](https://publicai.io/model-index/m/deepseek-v4)