- Context
- 1Mtokens
- Released
- 8mo ago
Benchmarks
DeepSWE is the primary coding benchmark where an exact model evaluation exists. Arena leaderboards and other sourced evaluations appear below; missing scores are not zero.
How it ranks
Against its predecessor, similarly priced models and the models ranked around it.
DeepSWE 1.1
Best pass@1 per model · Higher is better
Not evaluated in DeepSWE 1.1
Arena Text
Best Arena rating per model · Higher is better
Not ranked on Arena Text yet
Arena WebDev
Best Arena rating per model · Higher is better
Not ranked on Arena WebDev yet
Arena Vision
Best Arena rating per model · Higher is better
Not ranked on Arena Vision yet
Arena Search
Best Arena rating per model · Higher is better
Not ranked on Arena Search yet
Agent Arena
Best IPS score per model · Higher is better
Not ranked on Agent Arena yet
FrontierMath-2025-02-28-Private
Best score per model · Higher is better
#23 of 58
- 36.9%
Gemini 3.1 Pro PreviewSimilar release date - 32.4%
Claude Sonnet 4.6Similar release date - 23.1%
Qwen3.6 Max PreviewNearby rank - 21.4%
Kimi K2 ThinkingNearby rank - 21.0%Qwen3.5 Plus
- 20.7%Claude Opus 4.5Nearby rank
- 18.7%
o3Nearby rank - 16.4%
GLM 5Similar release date - 6.2%
Qwen3.5-FlashSimilar release date
Ahead of 2 of 4 models released around the same time
FrontierMath-Tier-4-2025-07-01-Private
Best score per model · Higher is better
#32 of 46
- 16.7%Gemini 3.1 Pro PreviewSimilar release date
- 8.3%Claude Sonnet 4.6Similar release date
- 2.1%GLM 4.6Nearby rank
- 2.1%GLM 5Similar release date
- 2.1%Qwen3.5 Plus
- 2.1%Claude Haiku 4.5Nearby rank
- 2.1%Gemini 2.5 Pro Preview 06-05Nearby rank
- 0.0%Qwen3.5-FlashSimilar release date
Ahead of 1 of 4 models released around the same time
GPQA diamond
Best score per model · Higher is better
#71 of 161
- 94.4%Gemini 3.1 Pro PreviewSimilar release date
- 87.8%GLM 5Similar release date
- 87.4%Claude Sonnet 4.6Similar release date
- 86.4%Qwen3.5 397B A17BSimilar release date
- 85.4%
Nemotron 3 UltraNearby rank - 85.3%Gemini 2.5 ProNearby rank
- 84.8%Qwen3.5 Plus
- 84.8%Gemini 2.5 Pro Preview 06-05Nearby rank
- 84.8%Qwen3.6 35B A3BNearby rank
- 56.8%Qwen3 8BPredecessor
+28.1 pts vs Qwen3 8B (predecessor) · Ahead of 0 of 4 models released around the same time
OTIS Mock AIME 2024-2025
Best score per model · Higher is better
#66 of 158
- 95.6%Gemini 3.1 Pro PreviewSimilar release date
- 88.9%Qwen3.5 397B A17BSimilar release date
- 87.8%
DeepSeek ReasonerNearby rank - 87.8%GPT-5.4 NanoNearby rank
- 86.7%Qwen3.5 Plus
- 86.7%Nemotron 3 UltraNearby rank
- 86.7%GPT-5 MiniNearby rank
- 85.8%Claude Sonnet 4.6Similar release date
- 80.0%GLM 5Similar release date
- 56.1%Qwen3 8BPredecessor
+30.6 pts vs Qwen3 8B (predecessor) · Ahead of 2 of 4 models released around the same time
SimpleQA Verified
Best score per model · Higher is better
#71 of 87
- 73.5%Gemini 3.1 Pro PreviewSimilar release date
- 46.3%GPT-5.4 ProSimilar release date
- 35.5%Claude Sonnet 4.6Similar release date
- 29.4%GPT-5.4 MiniNearby rank
- 26.0%GPT-4o (2024-08-06)Nearby rank
- 25.4%Qwen3.5 Plus
- 23.8%Claude Haiku 5.5Nearby rank
- 21.6%GPT-5 MiniNearby rank
- 20.3%Qwen3.5-FlashSimilar release date
Ahead of 1 of 4 models released around the same time
Not evaluated in DeepSWE 1.1 as of 2026-09-22. Results are only shown for exact model versions in the official dataset. Official leaderboard.
All published scores
- FrontierMath-2025-02-28-Private#23 of 5821.0%
- default21.0%
Source:Epoch AI evaluation - FrontierMath-Tier-4-2025-07-01-Private#32 of 462.1%
- default2.1%
Source:Epoch AI evaluation - GPQA diamond#71 of 16184.8%
+28.1 pts vs Qwen3 8B (predecessor)
- default84.8%
Source:Epoch AI evaluation - OTIS Mock AIME 2024-2025#66 of 15886.7%
+30.6 pts vs Qwen3 8B (predecessor)
- default86.7%
Source:Epoch AI evaluation - SimpleQA Verified#71 of 8725.4%
- default25.4%
Source:Epoch AI evaluation
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.