The LLM benchmark aggregator.

DeepSeek V3.1

DeepSeek · 685B · open weights

Strongest in Factual grounding (#18 of 104), weakest in Jailbreak resistance (#217 of 268). Above par in 6 of 16 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 4.8 and Grok 4.5 and ahead of Llama 3.1 405B Instruct and Claude 3.5 Haiku.

§ 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.

Core abilities50.2−15.1#96/1991/4

General intelligence50.2−19.2#85/1781/3

Human preference58.2−10.6#113/3541/1

Reasoning46.7−18.7#125/1951/5

Reasoning44.7−24.3#91/1471/3

Professional46.1−16.3#146/1941/2

Legal44.6−21.3#122/1731/1

Finance44.1−18.3#128/1621/1

Safety49.4−12#199/3292/3

Factual grounding61.3−9.4#18/1041/1

Fairness49.4−20.6#119/2921/2

Secure code52.5−13.7#132/2701/1

Harm refusal47.2−14.3#200/2921/2

Toxicity avoidance49.4−9.6#201/2681/1

Jailbreak resistance39.9−25.1#217/2681/1

§ 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 V3.1, left for the other.

Command R+won 12 · lost 1

Magistral Mediumwon 11 · lost 1

Jamba 1.6 Miniwon 9 · lost 1

Magistral Smallwon 9 · lost 1

GPT-3.5 Turbowon 9 · lost 1

Jamba 1.5 Miniwon 9 · lost 2

Magistral Small 1.2won 8 · lost 2

Mixtral 8x7B Instructwon 8 · lost 2

Llama 4 Scout Instructwon 12 · lost 4

DeepSeek V3won 12 · lost 4

Command Awon 11 · lost 4

Jamba 1.5 Largewon 8 · lost 3

Mistral Large 3won 8 · lost 3

Jamba 1.6 Largewon 7 · lost 3

Gemma 2 9B ITwon 7 · lost 3

Llama 3.1 70B Instructwon 7 · lost 3

Gemini 1.5 Flashwon 8 · lost 4

GPT-4.1 nanowon 10 · lost 5

GPT-4o miniwon 9 · lost 5

Gemma 3 4B ITwon 7 · lost 4

Gemma 3 27B ITwon 7 · lost 4

GPT-5.4 nanowon 7 · lost 4

Grok 3won 10 · lost 6

Gemini 1.5 Prowon 8 · lost 5

Llama 3.1 8B Instructwon 6 · lost 4

Gemini 2.0 Flashwon 9 · lost 6

Claude 3.5 Haikuwon 7 · lost 5

Llama 3.1 405B Instructwon 7 · lost 5

Llama 2 7B Chatwon 5 · lost 5

Llama 2 13B Chatwon 5 · lost 5

Llama 2 70B Chatwon 5 · lost 5

Claude 3 Haikuwon 5 · lost 5

Llama 3 70B Instructwon 5 · lost 5

GPT OSS 20Bwon 6 · lost 6

Llama 3.3 70B Instructwon 7 · lost 7

Llama 4 Maverick Instructwon 8 · lost 8

GPT OSS 120Bwon 8 · lost 8

DeepSeek R1won 8 · lost 8

Gemma 4 31Bwon 8 · lost 8

GPT-4.1 miniwon 7 · lost 8

Grok 4won 7 · lost 8

MiniMax M3won 7 · lost 8

DeepSeek V4 Flashwon 7 · lost 8

Claude 3.7 Sonnetwon 6 · lost 7

Grok 4 Fastwon 5 · lost 6

Gemini 2.5 Flashwon 7 · lost 9

GPT-4.1won 7 · lost 9

Kimi K2won 7 · lost 9

DeepSeek V3.2won 7 · lost 9

Claude Haiku 4.5won 6 · lost 8

Grok 4.1 Fastwon 6 · lost 8

Gemini 2.5 Flash-Litewon 5 · lost 7

Ling 3.0 Flashwon 4 · lost 6

Claude 3 Opuswon 4 · lost 6

GLM-4.6won 6 · lost 9

NVIDIA Nemotron 3 Ultra 550B A55Bwon 6 · lost 9

GLM-5.3 Flashwon 6 · lost 9

O3 Miniwon 5 · lost 8

Grok 4.3won 5 · lost 8

GPT-5 nanowon 6 · lost 10

O1 Miniwon 4 · lost 7

MiniMax M2.5won 4 · lost 7

GPT-4owon 5 · lost 9

GLM-4.5won 5 · lost 9

MiMo V2.6 Flashwon 5 · lost 9

Kimi K2.6won 5 · lost 9

GPT-4 Turbowon 4 · lost 8

MiniMax M2won 4 · lost 8

Claude 3.5 Sonnetwon 4 · lost 8

GLM-5.2won 5 · lost 10

O4 Miniwon 5 · lost 11

DeepSeek V4 Prowon 5 · lost 11

MiniMax M2.7won 4 · lost 9

Claude Sonnet 4.6won 4 · lost 9

Gemma 2 27B ITwon 3 · lost 7

Qwen2.5 72B Instructwon 3 · lost 7

GPT-5.4 miniwon 3 · lost 7

Gemini 3.5 Flashwon 4 · lost 10

Gemini 3.6 Flashwon 4 · lost 10

Mistral Large 4won 4 · lost 11

GPT-6 Lunawon 4 · lost 11

Grok 3 Miniwon 3 · lost 9

MiMo V2.5 Prowon 3 · lost 9

GPT-5 miniwon 4 · lost 12

Claude Sonnet 4won 4 · lost 12

Gemini 2.5 Prowon 4 · lost 12

Kimi K2.5won 4 · lost 12

Qwen3.7 Maxwon 3 · lost 11

Gemini 3.1 Flash-Litewon 2 · lost 8

Inklingwon 3 · lost 12

Qwen3 235B A22B Instructwon 3 · lost 13

Claude Opus 4won 3 · lost 13

Claude Opus 4.5won 2 · lost 9

Gemini 4 Argonwon 2 · lost 9

O1won 2 · lost 11

Claude Opus 4.6won 2 · lost 12

GPT-5.6 Lunawon 2 · lost 13

GPT-6 Solwon 2 · lost 13

MiMo V2.6 Prowon 2 · lost 13

Claude Sonnet 4.5won 2 · lost 14

GPT-5.4won 2 · lost 14

GPT-5.5won 2 · lost 14

GPT-5.6 Solwon 2 · lost 14

Gemini 3.1 Prowon 1 · lost 9

GLM-5won 1 · lost 10

GPT-6 Astrawon 1 · lost 10

Kimi K2 Thinkingwon 1 · lost 12

Gemini 3 Prowon 1 · lost 12

Muse Spark 1.1won 1 · lost 12

GPT-5.1won 1 · lost 13

Claude Sonnet 5won 1 · lost 13

Grok 4.6won 1 · lost 14

GPT-5.6 Terrawon 1 · lost 14

Qwen3.8 Maxwon 1 · lost 14

Gemini 3.8 Flashwon 1 · lost 14

GPT-5won 1 · lost 15

GPT-5.2won 1 · lost 15

Claude Opus 4.7won 1 · lost 15

Claude Opus 5.5won 1 · lost 15

Qwen3.5 397B A17Bwon 0 · lost 10

DeepSeek V4.1 Flashwon 0 · lost 11

GLM-5.1won 0 · lost 13

Claude Opus 5won 0 · lost 14

O3won 0 · lost 15

Grok 4.7won 0 · lost 15

Grok 4.5won 0 · lost 15

Claude Opus 4.8won 0 · lost 15

§ 3 · Sources

Where the numbers come from

6 publications, 12 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-11-10

LMArena Text 1416

Kagi LLM Benchmark ↗1 measureread 2026-11-10

Kagi LLM Benchmark 53.2%

SimpleBench ↗1 measureread 2026-11-10

SimpleBench 40%

Vals.ai ↗2 measuresread 2026-11-10

Vals · CaseLaw 53.91%Vals · CorpFin 51.48%

Enkrypt AI Safety Leaderboard ↗6 measuresread 2026-11-10

Enkrypt · Jailbreak risk 21.3%Enkrypt · Harmful content risk 41.1%Enkrypt · CBRN risk 18.3%Enkrypt · Toxicity risk 6.6%Enkrypt · Bias risk 78.8%Enkrypt · Insecure code risk 28%

Vectara Hallucination Leaderboard ↗1 measureread 2026-11-10

Vectara · Factual consistency 94.5%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=deepseek-v3-1)](https://publicai.io/model-index/m/deepseek-v3-1)