[whatmodel.sh](../../README.md) / Gemini 3.1 Pro (Preview)
# Gemini 3.1 Pro (Preview)
Google · released 2026-02 · closed weights
**score 73.8** ±0.4 · #41 of 732 · from 63 benchmark results across 29 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Gemini 3.1 Pro (Preview) among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆█▄▅▅●▄▂▁ #56 of 313 15% 10
coding ▁▁▁▃▅▇█▇▅●▃▁ #69 of 469 7.1% 4
chat & writing ▁▁▁▃▆▄▆█▆▇▇● #16 of 402* 2.3% 2
math ▁▁▃▅█▆▄█●▂▁▁ #33 of 431 44% 20
vision & documents ▁▁▂▂▁▃▄▂▆▇●▄ #21 of 153 6.2% 7
web search ▁▁▁▅●█▃▄·▁·▁ #53 of 84 2.1% 3
agent tasks · #56 of 313 · 15% of its score
Trails [Beam 1.0 501B A23B](beam-1-0-501b-a23b.md), [Claude Sonnet 4.6](claude-sonnet-4-6.md), [Qwen3.8 27B](qwen3-8-27b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena Agent (October 2026 aggrega… -0.0445 IPS ▁▁▁▃●▂█▁▂▂▁▁ beats 16/59 avg -0.379 IPS 5.3% sharp
Agents' Last Exam overall full-pa… 14.1% ▄···█▄▄▄●▄·█ beats 7/10 very hard · avg 1.9% 1.7% typical
τ²-bench Telecom (maker) 99.3% ▁·▅█▆▅▂▃▃▂▃● beats 88/89 medium · avg 61% 1.4% sharp · maxed out · small run
τ³-bench banking_knowledge (v1.0.… 26.0% ▃▃▂▃·●▃▂█▅▅▂ beats 10/25 very hard · avg 4.3% 1.4% at its level
MCP-Atlas (public set) 69.2% ●··█·██·████ beats 0/7 hard · avg 30% 1.1% typical
Vibe Code Bench (test split, 50 w… 32.0% ▃··▃·▃██●▃▃█ beats 9/14 very hard · avg 2.8% 1.0% sharp · at its level · noisy
SWE Atlas (Codebase Q&A, Test Wri… 23.7% ▄·▄··●···▄·█ beats 2/6 very hard · avg 6.6% 0.9% at its level
RWSearch 76.3% ●███··█····█ beats 0/5 medium · avg 45% 0.7% typical
Legal Research Bench (agentic, al… 20.8% ▄█··●▄▄██·▄█ beats 3/12 very hard · avg 6.3% 0.7% at its level
TerminalWorld-Verified (200 tasks… 55.0% ▄··▄·●·█···▄ beats 2/5 hard · avg 28% 0.4% blunt · at its level
coding · #69 of 469 · 7.1% of its score
Trails [MiMo-V2.5-Pro (1T-A42B)](mimo-v2-5-pro.md), [Qwen3.6 Plus](qwen3-6-plus.md), [MiniMax-M3](minimax-m3.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena WebDev (Code Arena) 1446.6 ▁▁▃▂▅█●▆▅▅▁▂ beats 62/129 avg 981 3.7% sharp
Luau Leetcode fully solved (Torpe… 82.7% ▁▁▁▁·▁▂▃▅▃█● beats 40/49 hard · avg 17% 1.5% sharp · at its level
Vibe Code Bench (test split, 50 w… 32.0% ▃··▃·▃██●▃▃█ beats 9/14 very hard · avg 2.8% 1.0% sharp · at its level · noisy
SWE Atlas (Codebase Q&A, Test Wri… 23.7% ▄·▄··●···▄·█ beats 2/6 very hard · avg 6.6% 0.9% at its level
chat & writing · #16 of 402 · 2.3% of its score
Beats [GPT-6 Astra](gpt-6-astra.md), [GPT-6.1 Sol](gpt-6-1-sol.md), [Claude Sonnet 5.5](claude-sonnet-5-5.md) here, though they rank higher overall. Trails [Muse Spark 1.2](muse-spark-1-2.md), [Muse Spark 1.1](muse-spark-1-1.md), [Claude Opus 4.6](claude-opus-4-6.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1486.9 ▁▁▁▂▃▃▃▆▆▇█● beats 369/385 avg 1260 2.1% blunt
IFEval (aggregation unspecified) 96.1% █·██·█···█·● beats 5/5 easy · avg 72% 0.3% maxed out
math · #33 of 431 · 44% of its score
Beats [Gemini 3.8 Flash](gemini-3-8-flash.md), [DeepSeek V4.1 Flash](deepseek-v4-1-flash.md), [DeepSeek V4 Pro (0813)](deepseek-v4-pro.md) here, though they rank higher overall. Trails [Qwen3.7 Max](qwen3-7-max.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
FrontierMath tiers 1–3 v2 59.6% ▂··▁▁▃▄▄●▆▄▃ beats 47/86 very hard · avg 1.2% 9.3% sharp · at its level
FrontierMath tier 4 v2 26.8% ▁▂▁▂●▅▂▃▂▁▁▁ beats 31/67 very hard · avg 0.1% 8.3% sharp · at its level
MathArena ArXivMath June 2026 (48… 66.7% ▂···▄·█●▆▆▄▂ beats 9/19 very hard · avg 5.1% 4.6% sharp · at its level
MathArena ArXivMath April 2026 (4… 63.7% ▁·····▅█●▄·▁ beats 15/20 very hard · avg 7.6% 4.3% sharp · at its level
MathArena ArXivMath February 2026… 62.5% ▁▁▁▁█▇▄▃▁●·▃ beats 22/24 very hard · avg 9.4% 2.5% at its level
MathArena ArXivMath March 2026 (3… 68.3% █▃▅·█▃▃·●·▃▃ beats 12/14 hard · avg 16% 1.9% at its level
MathArena ArXivMath January 2026 … 70.7% ▂··▄▂▄█▆▆▄●█ beats 18/23 hard · avg 23% 1.6% at its level
MathArena BrokenArXiv April 2026 … 25.8% ▂▂▆●▃▃▂▂▂··▂ beats 10/18 very hard · avg 0.8% 1.5% sharp · at its level · noisy
MathArena ArXivMath May 2026 (40 … 51.7% ▁··▁▄●▄▁·▃·▁ beats 5/17 very hard · avg 11% 1.5% at its level
MathArena ArXivMath December 2025… 66.2% ▃·▃▃█▁·▄·▁▁● beats 18/18 hard · avg 19% 1.2% at its level
MathArena ArXivLean June 2026 (46… 10.9% ▄····▄●███·▄ beats 2/9 very hard · avg 0.9% 1.1% sharp
MathArena BrokenArXiv June 2026 (… 17.6% ▄··█●▂▂▆▄··▂ beats 6/14 very hard · avg 1.3% 1.1% sharp · at its level · noisy
MathArena BrokenArXiv May 2026 (5… 17.0% ██·●▄▄····▄▄ beats 4/8 very hard · avg 1.9% 1.0% sharp
MathArena BrokenArXiv March 2026 … 13.8% █·▆●▂·▂▂···▂ beats 7/14 very hard · avg 1.0% 0.9% sharp · noisy
MathArena BrokenArXiv February 20… 18.5% ▃▂▂▃█●▂·▃··▂ beats 12/16 very hard · avg 0.9% 0.8% sharp · noisy
IMO-AnswerBench 90.0% ▁▁·▁▁▁▄▆█▄●▃ beats 40/44 hard · avg 25% 0.8% typical
MathArena ArXivLean March 2026 (4… 14.6% ▂····▂·····● beats 2/5 very hard · avg 1.2% 0.7% typical
MathNet-Solve-Test micro-average … 78.4% ▃▃···█▃·▃·▃● beats 8/8 very hard · avg 15% 0.6% at its level
MathConstraint full release, no t… 51.7% ▄··▄▄·▄▄█●·▄ beats 7/8 very hard · avg 11% 0.5% at its level
OTIS mock AIME 2024–2025 95.6% ▃▃▄▃▃▃▇▇█●▂▅ beats 164/197 very hard · avg 7.2% 0.3% sharp · maxed out · noisy
vision & documents · #21 of 153 · 6.2% of its score
Beats [Claude Sonnet 5.5](claude-sonnet-5-5.md), [GPT-6 Sol](gpt-6-sol.md), [GPT-5.5](gpt-5-5.md) here, though they rank higher overall. Trails [Gemini 3.6 Flash](gemini-3-6-flash.md), [Gemini 3.5 Flash](gemini-3-5-flash.md), [Muse Spark 1.2](muse-spark-1-2.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
CharXiv (reasoning questions) (ma… 83.3% ▃·▃·▃▃▃·▃▃●▅ beats 7/11 medium · avg 38% 2.1% typical
Arena vision (style controlled) 1279.5 ▁▁▂▂▂▂▃▄▅▅█● beats 130/152 avg 1061 1.4% blunt
ChartArena chart parsing (English… 59.2% ▁·▁·▃▃▁▇▃█·● beats 22/22 hard · avg 27% 1.1% blunt · at its level
Arena document (style controlled) 1458.9 ▁▁▃▃▄▃▇●▅▇▄▅ beats 16/37 avg 1331 0.7% blunt
BEAR-Bench (bilingual RU/EN enter… 75.1% ▃▃·▅·▃·█·█▅● beats 13/14 hard · avg 24% 0.6% at its level
Diagram-MMU diagram question answ… 86.3% ▄·····▄▄·█·● beats 5/6 medium · avg 56% 0.3% blunt
PureDocBench document parsing (Av… 70.4% ▂·▁·▂▂▅●·▁▂▂ beats 16/22 medium · avg 62% 0.0% blunt · at its level
web search · #53 of 84 · 2.1% of its score
Trails [GPT-4o mini (2024-07-18)](gpt-4o-mini-2024-07-18.md), [Mistral Small 3.1 24B (2503)](mistral-small-3-1.md), [Llama 4 Scout (109B-A17B)](llama-4-scout.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena search systems (style contr… 1206.6 ▁·▁▁▂▂▂█●▃▂▂ beats 25/36 avg 1069 0.7% blunt
RWSearch 76.3% ●███··█····█ beats 0/5 medium · avg 45% 0.7% typical
Legal Research Bench (agentic, al… 20.8% ▄█··●▄▄██·▄█ beats 3/12 very hard · avg 6.3% 0.7% at its level
other tests (21)
test result vs the others in its run difficulty counts why
EBR-bench (Epoch) 14.3% ▄▂·▆●▂█▆▄▆·▆ beats 8/25 very hard · avg 0.3% 3.5% sharp
Furniture Assembly (Epoch) 26.7% █●▆▆▆▂▆▂▂▂▂▅ beats 5/32 very hard · avg 1.2% 3.4% sharp · at its level
Mystery Game Puzzles (Epoch) 34.0% █··▂▃▇▄●▂▄▂▁ beats 54/77 very hard · avg 9.4% 3.1% sharp · noisy
Chess Puzzles (Epoch) 49.0% █▁▁▁▁▂▃▃▃●▁▁ beats 130/137 very hard · avg 5.8% 1.9% sharp · at its level · noisy
KINA (Knowledge Index of Noah's A… 53.2% ▁▁▂▁▃▄█▆▄▅▂● beats 39/39 very hard · avg 21% 1.7% at its level
GPQA Diamond 94.4% ▁▁▁▁▂▄▆▄▅▇█● beats 205/211 hard · avg 39% 1.5% sharp
MMLU-Pro 91.1% ▃▂▅█▃·█▃▅··● beats 23/23 medium · avg 54% 1.3% typical
EQ-Bench 3 (emotional intelligenc… 1301.8 ▁·▂▁▄▅▅▃●▄▁▁ beats 45/62 avg 1034 1.3% typical
The Token Games Elo (10-model puz… 1129.0 █··▄█··▄·●·▄ beats 6/7 avg 822 1.1% typical
WildClawBench 40.8% ▃·▅▅█·●█▅▃▃▃ beats 9/17 very hard · avg 11% 1.1% at its level
AgentWorldBench (open-ended, over… 54.6% ▄▄··▄▄·▄●▄██ beats 5/11 hard · avg 26% 1.0% blunt · at its level
Legal Research Bench (single-shot… 6.5% ▃▃█▃·▅···●▃▃ beats 9/12 very hard · avg 0.8% 0.9% too hard for it
SimpleQA Verified 73.5% ▁▁▃▂▃▃▆▆█▄▂● beats 79/81 very hard · avg 8.0% 0.8% at its level
Icelandic LLM Leaderboard average… 66.5% ▁▁▃▂▂▄▁▃▄▄▄● beats 33/38 very hard · avg 14% 0.8% at its level
FORTE (AGI-Eval) 70.3% ●···█·····▄█ beats 0/5 medium · avg 38% 0.7% at its level
WritingBench 83.7% ▁·▁▂▁··●█▂▂▁ beats 7/20 medium · avg 54% 0.7% typical
MirrorCode (Epoch) 8.9% ▄·●▄▄▄██·▄▄█ beats 1/13 very hard · avg 1.6% 0.5% sharp · at its level · noisy
RobloxQA v2.0 multiple choice (To… 96.5% ▁▁▃▁█▄▃▃●▂▂▂ beats 36/49 easy · avg 77% 0.5% maxed out
FinSheet-Bench accuracy (501 ques… 82.4% █···██·····● beats 3/3 medium · avg 53% 0.4% typical
DiagnosticIQ industrial maintenan… 72.7% ▁··▃▁▁▁▅█▅▁● beats 22/24 hard · avg 51% 0.4% blunt · at its level
Vectara hallucination leaderboard… 89.6% ▁▂▁▂▅●▂▃▄▃▁▁ beats 42/98 very easy · avg 90% 0.0% blunt
source results share of the score
Epoch AI 10 33% ▇▇▇▇
MathArena ArXivMath / BrokenArXiv / ArX… 14 24% ▇▇▇
LMArena leaderboard dataset 6 14% ▇▇
Google DeepMind Gemini model page and m… 2 4% ▇
meituan-longcat/LongCat-2.0 model card 5 3% ▇
Roblox LLM Leaderboard 2 2% ▇
Knowledge Index of Noah's Ark 1 2% ▇
Agents' Last Exam 1 2% ▇
Legal Research Bench 2 2% ▇
τ²/τ³-bench leaderboard: Sierra's own e… 1 1% ▇
MMLU-Pro author-released evaluation runs 1 1% ▇
EQ-Bench 3 canonical leaderboard results 1 1% ▇
ChartArena: Benchmarking Chart Parsing … 1 1% ▇
The Token Games: Evaluating Language Mo… 1 1% ▇
WildClawBench: A Benchmark for Real-Wor… 1 1% ▇
zai-org/GLM-5.2 model card 1 1% ▇
Qwen/Qwen-AgentWorld-35B-A3B model card 1 1% ▇
Vibe Code Bench 1 1% ▇
SWE Atlas: Benchmarking Coding Agents B… 1 1% ▇
Miðeind Icelandic LLM Leaderboard 1 1% ▇
MathNet: a Global Multimodal Benchmark … 1 1% ▇
BEAR-Bench: A Bilingual Enterprise and … 1 1% ▇
MathConstraint: Automated Generation of… 1 0% ▇
FinSheet-Bench: From Simple Lookups to … 1 0% ▇
DiagnosticIQ: A Benchmark for LLM-Based… 1 0% ▇
TerminalWorld: Benchmarking Agents on R… 1 0% ▇
Diagram-MMU: A Multi-Modal Benchmark fo… 1 0% ▇
How Far Is Document Parsing from Solved… 1 0% ▇
Vectara hallucination leaderboard 1 0% ▇
prior: pull toward the middle 0% ▇
weighted average = 73.8
Licences: Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · MathArena ArXivMath / BrokenArXiv / ArXivLean model-output datasets ([CC BY-SA 4.0](../../licenses/matharena-sharealike-NOTICE.txt)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · meituan-longcat/LongCat-2.0 model card ([mit](../../licenses/meituan-longcat--LongCat-2.0.txt)) · Roblox LLM Leaderboard ([MIT](../../licenses/roblox-llm-leaderboard-MIT.txt)) · Knowledge Index of Noah's Ark ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Agents' Last Exam ([CC BY 4.0](../../licenses/paper-2606.05405v2-CC-BY-4.0.txt)) · Legal Research Bench ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · τ²/τ³-bench leaderboard: Sierra's own evaluation runs ([MIT](https://opensource.org/license/mit)) · MMLU-Pro author-released evaluation runs ([Apache-2.0](../../licenses/mmlu-pro-author-runs-Apache-2.0.txt)) · EQ-Bench 3 canonical leaderboard results ([MIT](../../licenses/eqbench3-MIT.txt)) · ChartArena: Benchmarking Chart Parsing across Languages ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · The Token Games: Evaluating Language Model Reasoning with Puzzle Duels ([CC BY-SA 4.0](../../licenses/paper-2602.17831v2-CC-BY-SA-4.0.txt)) · WildClawBench: A Benchmark for Real-World ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · zai-org/GLM-5.2 model card ([mit](../../licenses/zai-org--GLM-5.2.txt)) · Qwen/Qwen-AgentWorld-35B-A3B model card ([Apache-2.0](https://huggingface.co/Qwen/Qwen-AgentWorld-35B-A3B/blob/60d2b0434a53d2e62a7c00a489586815d94ebffb/README.md)) · Vibe Code Bench ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · SWE Atlas: Benchmarking Coding Agents Beyond Issue Resolution ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Miðeind Icelandic LLM Leaderboard ([CC BY 4.0](../../licenses/mideind-icelandic-llm-leaderboard-CC-BY-4.0.txt)) · MathNet: a Global Multimodal Benchmark for Mathematical Reasoning and Retrieval ([CC BY 4.0](../../licenses/paper-2604.18584v2-CC-BY-4.0.txt)) · BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · MathConstraint: Automated Generation of Verified Combinatorial Reasoning Instances for LLMs ([CC BY 4.0](../../licenses/paper-2605.08498v1-CC-BY-4.0.txt)) · FinSheet-Bench: From Simple Lookups to Complex Reasoning ([CC BY 4.0](../../licenses/paper-2603.07316v1-CC-BY-4.0.txt)) · DiagnosticIQ: A Benchmark for LLM-Based Industrial Maintenance Action Recommendation from Symbolic Rules ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · TerminalWorld: Benchmarking Agents on Real-World Terminal Tasks ([CC BY 4.0](../../licenses/paper-2605.22535v1-CC-BY-4.0.txt)) · Diagram-MMU: A Multi-Modal Benchmark for Scientific Diagrams ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · How Far Is Document Parsing from Solved? PureDocBench: A Source-Traceable Benchmark across Clean ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Vectara hallucination leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0))
## How to run it
### Through an API
api (Google) · $4.50/M blended · $2 in / $12 out per 1M tokens · [ai.google.dev](https://ai.google.dev/gemini-api/docs/models)
[price source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/providers/google/models/gemini-3.1-pro-preview.toml): Standard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ. · [context source](https://ai.google.dev/gemini-api/docs/models/gemini-3.1-pro-preview): Gemini Developer API, gemini-3.1-pro-preview: maximum input tokens; output token limit is separate.
[interactive version](https://whatmodel.sh/model/gemini-3-1-pro) · [JSON](https://whatmodel.sh/api?format=json&all=1)