- input
- $0.45
- cached
- $0.10
- output
- $1.20
- Context
- 524Ktokens
- Released
- 3mo ago
- Size
- 266Bparameters
Cost/1M tokens
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
Arena #160
- 1413
Qwen3 VL 235B A22B InstructSimilar price - 1409Qwen3.5-27BNearby rank
- 1408
Mistral Medium 3.1Nearby rank - 1405
Inkling Small - 1405
X-Ai/Grok-4-Fast-ReasoningNearby rank - 1402
Qwen3 235B-A22BNearby rank - 1390
GPT-5 MiniSimilar price - 1318
Llama 3.3 70B InstructSimilar price - 1290
Gemma 2 27BSimilar price
Ahead of 3 of 4 similarly priced models
Arena WebDev
Best Arena rating per model · Higher is better
Arena #99
- 1624
MiMo-V2.6-ProSimilar price - 1590Qwen3.8 27BSimilar price
- 1417GPT-5Nearby rank
- 1414InklingNearby rank
- 1408Inkling Small
- 1406GPT-5.3-CodexNearby rank
- 1400Qwen3.5 397B A17BNearby rank
- 1256Gemini 3.1 Flash Lite PreviewSimilar price
- 1245GPT-5.1-Codex-MiniSimilar price
Ahead of 2 of 4 similarly priced models
Arena Vision
Best Arena rating per model · Higher is better
Arena #89
- 1247MiMo-V2.6-ProSimilar price
- 1235Gemini 3.1 Flash Lite PreviewSimilar price
- 1214Qwen3 VL 235B A22B InstructSimilar price
- 1211GPT-5Nearby rank
- 1208
Claude Opus 4Nearby rank - 1208Inkling Small
- 1206Claude Sonnet 4Nearby rank
- 1201GPT-4.1 MiniNearby rank
- 1185GPT-5 MiniSimilar price
Ahead of 1 of 4 similarly priced models
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
Arena #58
- +0.053
Hy4Similar price - +0.024MiMo-V2.6-ProSimilar price
- -0.048Qwen3.8 27BSimilar price
- -0.105
MiniMax M3Similar price - -0.156
Nemotron 3 UltraNearby rank - -0.173InklingNearby rank
- -0.181Inkling Small
- -0.226
Solar Pro 4Nearby rank
Ahead of 0 of 4 similarly priced models
ARC-AGI-2
Best score per model · Higher is better
#44 of 85
- 72.9%
DeepSeek V4.1 FlashSimilar price - 52.6%Grok 4.5Nearby rank
- 42.4%Qwen3.8 27BSimilar price
- 40.1%Inkling Small
- 37.6%Claude Opus 4.5Nearby rank
- 36.5%InklingNearby rank
- 4.4%GPT-5 MiniSimilar price
- 0.0%Llama 4 MaverickSimilar price
Ahead of 2 of 4 similarly priced models
FrontierMath-Tier-4-v2-Private
Best score per model · Higher is better
#51 of 72
- 29.3%
GLM 5.3Similar price - 26.8%DeepSeek V4.1 FlashSimilar price
- 19.5%GPT-5 ProNearby rank
- 17.1%Gemini 3 Flash PreviewNearby rank
- 17.1%Inkling Small
- 17.1%Grok 4.7Nearby rank
- 17.1%Grok 4.20 (Reasoning)Nearby rank
- 12.2%GPT-5 MiniSimilar price
- 0.0%Gemini 3.5 Flash LiteSimilar price
Ahead of 2 of 4 similarly priced models
FrontierMath-Tiers-1-3-v2-Private
Best score per model · Higher is better
#56 of 87
- 67.4%DeepSeek V4.1 FlashSimilar price
- 51.2%GPT-5.4 MiniNearby rank
- 46.7%GPT-5 MiniSimilar price
- 46.3%Inkling Small
- 45.3%DeepSeek V4 Pro 0423Nearby rank
- 44.9%GPT-5.4 NanoNearby rank
- 34.4%Qwen3.7 PlusSimilar price
- 27.7%Gemini 3.1 Flash LiteSimilar price
Ahead of 2 of 4 similarly priced models
GPQA diamond
Best score per model · Higher is better
#51 of 161
- 89.3%Grok 4.20 (Reasoning)Nearby rank
- 88.8%Grok 4.3Nearby rank
- 88.5%Inkling Small
- 88.4%Qwen3.6 PlusNearby rank
- 88.3%InklingNearby rank
- 87.9%Qwen3.7 PlusSimilar price
- 75.0%GPT-5 MiniSimilar price
- 47.4%Llama 3.3 70B InstructSimilar price
- 36.5%Gemma 2 27BSimilar price
Ahead of 4 of 4 similarly priced models
OTIS Mock AIME 2024-2025
Best score per model · Higher is better
#58 of 158
- 93.3%Qwen3.7 PlusSimilar price
- 91.1%Qwen3.6 27BNearby rank
- 91.1%GLM 5.3Nearby rank
- 90.0%Inkling Small
- 88.9%GPT-5.4 MiniNearby rank
- 88.9%gpt-oss-120bNearby rank
- 86.7%GPT-5 MiniSimilar price
- 5.1%Llama 3.3 70B InstructSimilar price
- 1.4%Gemma 2 27BSimilar price
Ahead of 3 of 4 similarly priced models
SimpleQA Verified
Best score per model · Higher is better
#76 of 87
- 44.1%Qwen3.6 PlusSimilar price
- 42.9%DeepSeek V4.1 FlashSimilar price
- 34.3%
Kimi K2.5Similar price - 21.6%GPT-5 MiniSimilar price
- 20.3%Qwen3.5-FlashNearby rank
- 20.0%Mistral Large 4Nearby rank
- 19.1%Inkling Small
- 17.4%
Seed 1.6Nearby rank - 15.9%Qwen3.6 FlashNearby rank
Ahead of 0 of 4 similarly priced models
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.
Reasoning effort
ARC-AGI-2
Score by reasoning effort · Higher is better
All published scores
- Arena TextArena #1601405
- default140595% CI 1401 to 1410 · 37,665 votes
- Arena WebDevArena #991408
- default140895% CI 1398 to 1417 · 5,220 votes
- Arena VisionArena #891208
- default120895% CI 1200 to 1216 · 9,259 votes
- Agent ArenaArena #58-0.181
- default-0.18195% CI -0.201 to -0.160 · 568,442 observations
- ARC-AGI-2#44 of 8540.1%
- minimal4.0%
- none0.0%
- low4.9%
- medium13.6%
- high33.1%
- xhigh40.1%
- FrontierMath-Tier-4-v2-Private#51 of 7217.1%
- xhigh17.1%
Source:Epoch AI evaluation - FrontierMath-Tiers-1-3-v2-Private#56 of 8746.3%
- xhigh46.3%
Source:Epoch AI evaluation - GPQA diamond#51 of 16188.5%
- xhigh88.5%
Source:Epoch AI evaluation - OTIS Mock AIME 2024-2025#58 of 15890.0%
- xhigh90.0%
Source:Epoch AI evaluation - SimpleQA Verified#76 of 8719.1%
- xhigh19.1%
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.
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.