Decision support
Composer 1 vs Gemini 3 Pro Preview
Compare canonical objects directly. The useful dimensions change with the object type, so model and hardware comparisons stay separate.
Cursor
Benchmark results
Arena Text
Arena rating · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview1486Arena #21CI 1482 to 1489
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default148695% CI 1482 to 1489 · 41,829 votes
Arena WebDev
Arena rating · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview1441Arena #86CI 1432 to 1449
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default144195% CI 1432 to 1449 · 13,929 votes
Arena Vision
Arena rating · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview1289Arena #14CI 1282 to 1297
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default128995% CI 1282 to 1297 · 13,547 votes
Arena Search
Arena rating · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview1201Arena #18CI 1195 to 1206
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default120195% CI 1195 to 1206 · 38,324 votes
Arena Document
Arena rating · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview1451Arena #32CI 1442 to 1460
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default145195% CI 1442 to 1460 · 11,338 votes
ARC-AGI-2
Best published score · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview31.1%#48 of 85
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default31.1%
Source:External evaluation via Epoch AIOriginal-source terms applyFrontierMath-2025-02-28-Private
Best published score · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview37.6%#9 of 58
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default37.6%
Source:Epoch AI evaluationFrontierMath-Tier-4-2025-07-01-Private
Best published score · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview18.8%#6 of 46
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default18.8%
Source:Epoch AI evaluationGPQA diamond
Best published score · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview92.6%#25 of 161
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default92.6%
Source:Epoch AI evaluationHumanity's Last Exam
Best published score · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview37.5%#7 of 34
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default37.5%
Source:External evaluation via Epoch AIOriginal-source terms applyOTIS Mock AIME 2024-2025
Best published score · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview91.4%#52 of 158
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default91.4%
Source:Epoch AI evaluationSWE-bench Verified
Best published score · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview72.9%#23 of 35
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default72.9%
Source:Epoch AI evaluationTerminal Bench
Best published score · Higher is better
- Composer 1Not reported
- Gemini 3 Pro Preview69.4%#7 of 49
Runs and sources
Composer 1Not reported
Gemini 3 Pro Preview- default69.4%Ante
- default65.2%SageAgent
- default62.2%CodeBrain-1.5
- default62.2%CodeBrain-1
- default61.8%II-Agent
- default61.1%Droid
- default56.9%Terminus 2
- default56.0%Letta Code
Sources:External evaluation via Epoch AIOriginal-source terms applyhttps://www.tbench.ai/leaderboard/terminal-bench/2.0Original-source terms apply
Each model's best published result is shown; "Not reported" means no matched evaluation, not a score of zero. Arena and DeepSWE ranks are the leaderboards' own; other ranks are among catalogue models with a score. Results from different harnesses or settings are not directly comparable.
Original-source terms apply to evaluations credited to their original publishers.
Benchmark compilation: Epoch AI, Capabilities & benchmarking (CC BY attribution). Scores are converted to percentages and model names matched to the Vertical Tabs catalog. Individual evaluations are credited to their original sources. External results may carry separate licences and evaluation configurations; do not interpret scores from different harnesses as directly comparable.
Arena ratings: Arena (lmarena-ai/leaderboard-dataset), licensed CC BY 4.0. Ratings come from pairwise human votes and are only comparable within one leaderboard; the interval is Arena's 95% confidence interval. Model names are matched to the Vertical Tabs catalog.
Model specifications
API pricing · USD per 1M tokens
Capabilities and modalities
Architecture and availability
A dash means the field is not reported. Provider-specific performance and pricing may differ.