- Context
- 1Mtokens
- Released
- 8mo ago
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
#27 of 28
- 74.1%
- 73.8%
- 73.6%
- 72.7%
- 69.9%
- 69.6%
- 69.0%
- 68.5%
- 67.5%
- 67.2%
- 29.9%Claude Sonnet 4.6
Arena Text
Best Arena rating per model · Higher is better
Arena #48
- 1511
Muse WonderSimilar price - 1487Gemini 3.1 Pro PreviewSimilar price
- 1475GPT-5.4Similar price
- 1473GPT-5.5 InstantNearby rank
- 1472Claude Sonnet 5.5Nearby rank
- 1472Claude Sonnet 4.6
- 1468GPT-5.6 TerraNearby rank
- 1467
ERNIE 5.1Nearby rank - 1456Claude Sonnet 4.5Predecessor
- 1354
Command ASimilar price
+16 vs Claude Sonnet 4.5 (predecessor) · Ahead of 1 of 4 similarly priced models
Arena WebDev
Best Arena rating per model · Higher is better
Arena #60
- 1658Kimi K3Similar price
- 1649Muse WonderSimilar price
- 1534Muse Spark 1.2Nearby rank
- 1526Muse Spark OpenNearby rank
- 1522Claude Sonnet 4.6
- 1520GPT-5.6 TerraNearby rank
- 1519GPT-5.6 LunaNearby rank
- 1462GPT-5.4Similar price
- 1446Gemini 3.1 Pro PreviewSimilar price
- 1393Claude Sonnet 4.5Predecessor
+129 vs Claude Sonnet 4.5 (predecessor) · Ahead of 2 of 4 similarly priced models
Arena Vision
Best Arena rating per model · Higher is better
Arena #34
- 1282GPT-5.4Similar price
- 1280Grok 4.5Similar price
- 1280Gemini 3.1 Pro PreviewSimilar price
- 1278GPT-5.2 ChatNearby rank
- 1277GPT-5.5 InstantNearby rank
- 1275Claude Sonnet 4.6
- 1272GLM 5.3 FlashNearby rank
- 1266GPT-5.6 TerraNearby rank
- 1206Claude Sonnet 4Earlier version
- 1199
Mistral Medium 3.5Similar price
+69 vs Claude Sonnet 4 (earlier version) · Ahead of 1 of 4 similarly priced models
Arena Search
Best Arena rating per model · Higher is better
Not ranked on Arena Search yet
Arena Document
Best Arena rating per model · Higher is better
Arena #13
- 1481GPT-5.5Nearby rank
- 1480Muse Spark 1.1Nearby rank
- 1478Claude Sonnet 4.6
- 1474GPT-5.6 TerraNearby rank
- 1474Muse Spark 1.3Nearby rank
- 1471GPT-5.4Similar price
- 1466Grok 4.5Similar price
- 1459Gemini 3.1 Pro PreviewSimilar price
- 1448Claude Sonnet 4.5Predecessor
- 1425GPT-5.2Similar price
+30 vs Claude Sonnet 4.5 (predecessor) · Ahead of 2 of 4 similarly priced models
Agent Arena
Best IPS score per model · Higher is better
Not ranked on Agent Arena yet
ARC-AGI-2
Best score per model · Higher is better
#35 of 85
- 77.1%Gemini 3.1 Pro PreviewSimilar price
- 74.0%GPT-5.4Similar price
- 67.1%Grok 4.6Similar price
- 61.3%Claude Haiku 5.5Nearby rank
- 61.3%
DeepSeek V4 Pro 0813Nearby rank - 60.4%Claude Sonnet 4.6
- 60.4%Gemini 3.6 FlashNearby rank
- 60.4%Kimi K3Similar price
- 13.6%Claude Sonnet 4.5Predecessor
+46.8 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 0 of 4 similarly priced models
FrontierMath-2025-02-28-Private
Best score per model · Higher is better
#14 of 58
- 47.6%GPT-5.4Similar price
- 40.7%GPT-5.2Similar price
- 36.9%Gemini 3.1 Pro PreviewSimilar price
- 33.4%GLM 5.1Nearby rank
- 32.4%GPT-5Nearby rank
- 32.4%Claude Sonnet 4.6
- 31.0%GPT-5.1Nearby rank
- 28.3%GPT-5.4 MiniNearby rank
- 15.2%Claude Sonnet 4.5Predecessor
- 0.3%GPT-4o (2024-08-06)Similar price
+17.2 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 1 of 4 similarly priced models
FrontierMath-Tier-4-2025-07-01-Private
Best score per model · Higher is better
#15 of 46
- 27.1%GPT-5.4Similar price
- 18.8%GPT-5.2Similar price
- 16.7%Gemini 3.1 Pro PreviewSimilar price
- 12.5%GLM 5.1Nearby rank
- 8.3%
Qwen3.6 PlusNearby rank - 8.3%Claude Sonnet 4.6
- 6.3%GPT-5 MiniNearby rank
- 6.3%GPT-5.4 NanoNearby rank
- 4.2%Claude Sonnet 4.5Predecessor
- 2.1%o3Similar price
+4.1 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 1 of 4 similarly priced models
GPQA diamond
Best score per model · Higher is better
#59 of 161
- 94.4%Gemini 3.1 Pro PreviewSimilar price
- 94.0%Grok 4.6Similar price
- 93.3%GPT-5.4Similar price
- 93.1%Kimi K3Similar price
- 87.6%GPT-5.1Nearby rank
- 87.4%
Qwen3.6 Max PreviewNearby rank - 87.4%Claude Sonnet 4.6
- 86.9%GPT-5.4 MiniNearby rank
- 86.4%Qwen3.5 397B A17BNearby rank
- 82.3%Claude Sonnet 4.5Predecessor
+5.1 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 0 of 4 similarly priced models
OSWorld
Best score per model · Higher is better
#1 of 10
- 72.1%Claude Sonnet 4.6
- 66.3%Claude Opus 4.5Similar price
- 63.3%Kimi K2.5Nearby rank
- 62.9%Claude Sonnet 4.5Predecessor
- 23.0%o3Similar price
+9.2 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 2 of 2 similarly priced models
OSWorld 2.0
Best score per model · Higher is better
#6 of 9
- 31.4%Claude Opus 5Similar price
- 27.3%GPT-5.6 SolSimilar price
- 18.5%Claude Opus 4.8Similar price
- 13.9%Claude Opus 4.7Nearby rank
- 13.0%GPT-5.5Similar price
- 9.3%Claude Sonnet 4.6
- 4.6%
MiniMax M3Nearby rank - 4.6%Kimi K2.6Nearby rank
Ahead of 0 of 4 similarly priced models
OTIS Mock AIME 2024-2025
Best score per model · Higher is better
#75 of 158
- 100.0%Qwen3.8 Max (0902)Similar price
- 97.8%GPT-5.4Similar price
- 97.2%Kimi K3Similar price
- 95.6%Gemini 3.1 Pro PreviewSimilar price
- 86.4%GLM 5.2Nearby rank
- 86.1%Claude Opus 4.5Nearby rank
- 85.8%Claude Sonnet 4.6
- 84.7%Gemini 2.5 ProNearby rank
- 84.4%GPT-5.5Nearby rank
- 77.8%Claude Sonnet 4.5Predecessor
+8.0 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 0 of 4 similarly priced models
SimpleQA Verified
Best score per model · Higher is better
#56 of 87
- 73.5%Gemini 3.1 Pro PreviewSimilar price
- 56.0%Grok 4.7Similar price
- 50.6%Kimi K3Similar price
- 45.1%GPT-5.4Similar price
- 37.1%GPT-5.2Nearby rank
- 36.5%Kimi K2.7 CodeNearby rank
- 35.5%Claude Sonnet 4.6
- 34.9%Kimi K2.6Nearby rank
- 34.3%Kimi K2.5Nearby rank
- 30.7%Claude Sonnet 4.5Predecessor
+4.8 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 0 of 4 similarly priced models
SWE-bench Verified
Best score per model · Higher is better
#15 of 35
- 76.9%GPT-5.4Similar price
- 75.6%Gemini 3.1 Pro Preview Custom ToolsSimilar price
- 75.4%Gemini 3 Flash PreviewNearby rank
- 75.2%Claude Sonnet 4.6
- 74.8%GPT-5.3-CodexSimilar price
- 74.2%GLM 5.1Nearby rank
- 73.8%GPT-5.2Similar price
- 71.3%Claude Sonnet 4.5Predecessor
+3.9 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 2 of 4 similarly priced models
Terminal Bench
Best score per model · Higher is better
#17 of 49
- 81.8%GPT-5.4Similar price
- 80.2%Gemini 3.1 Pro PreviewSimilar price
- 78.4%GPT-5.3-CodexSimilar price
- 66.5%GPT-5.2-CodexSimilar price
- 57.8%GPT-5.1-CodexNearby rank
- 57.3%Grok 4.20Nearby rank
- 53.4%Claude Sonnet 4.6
- 52.4%GLM 5Nearby rank
- 49.6%GPT-5Nearby rank
- 46.5%Claude Sonnet 4.5Predecessor
+6.9 pts vs Claude Sonnet 4.5 (predecessor) · Ahead of 0 of 4 similarly priced models
DeepSWE 1.1
Primary coding-agent benchmark · 113 tasks · Data generated 2026-09-22
- pass@4
- 56.6%
- Average cost / attempt
- $5.52
- Benchmark runs
- 4
pass@1 is the success rate across scored attempts; pass@4 is the share of tasks solved in at least one of four attempts. Costs are reported by DeepSWE for each configuration. These results are not directly comparable to evaluations using different harnesses. Official source JSON.
Reasoning effort
ARC-AGI-2
Score by reasoning effort · Higher is better
GPQA diamond
Score by reasoning effort · Higher is better
OSWorld 2.0
Score by reasoning effort · Higher is better
OTIS Mock AIME 2024-2025
Score by reasoning effort · Higher is better
SimpleQA Verified
Score by reasoning effort · Higher is better
All published scores
- Arena TextArena #481472
+16 vs Claude Sonnet 4.5 (predecessor)
- default147295% CI 1468 to 1476 · 70,017 votes
- Arena WebDevArena #601522
+129 vs Claude Sonnet 4.5 (predecessor)
- default152295% CI 1517 to 1527 · 22,548 votes
- Arena VisionArena #341275
+69 vs Claude Sonnet 4 (earlier version)
- default127595% CI 1269 to 1281 · 27,201 votes
- Arena DocumentArena #131478
+30 vs Claude Sonnet 4.5 (predecessor)
- default147895% CI 1472 to 1484 · 61,055 votes
- ARC-AGI-2#35 of 8560.4%
+46.8 pts vs Claude Sonnet 4.5 (predecessor)
- high60.4%
- max58.3%
- FrontierMath-2025-02-28-Private#14 of 5832.4%
+17.2 pts vs Claude Sonnet 4.5 (predecessor)
- default32.4%16K
Source:Epoch AI evaluation - FrontierMath-Tier-4-2025-07-01-Private#15 of 468.3%
+4.1 pts vs Claude Sonnet 4.5 (predecessor)
- default8.3%16K
Source:Epoch AI evaluation - GPQA diamond#59 of 16187.4%
+5.1 pts vs Claude Sonnet 4.5 (predecessor)
- medium83.3%
- high83.3%
- max78.8%
- default87.4%32K
Source:Epoch AI evaluation - OSWorld#1 of 1072.1%
+9.2 pts vs Claude Sonnet 4.5 (predecessor)
- default72.1%claude-sonnet-4-6 (100 steps)
- OSWorld 2.0#6 of 99.3%
- medium9.3%
- max8.3%
- OTIS Mock AIME 2024-2025#75 of 15885.8%
+8.0 pts vs Claude Sonnet 4.5 (predecessor)
- medium82.2%
- high75.6%
- max71.1%
- default85.8%32K
Source:Epoch AI evaluation - SimpleQA Verified#56 of 8735.5%
+4.8 pts vs Claude Sonnet 4.5 (predecessor)
- high35.5%
- max32.8%
Source:Epoch AI evaluation - SWE-bench Verified#15 of 3575.2%
+3.9 pts vs Claude Sonnet 4.5 (predecessor)
- default75.2%
Source:Epoch AI evaluation - Terminal Bench#17 of 4953.4%
+6.9 pts vs Claude Sonnet 4.5 (predecessor)
- unknown53.4%Simplai Agenthttps://www.tbench.ai/leaderboard/terminal-bench/2.0Original-source terms apply
- default53.4%Simplai AgentExternal evaluation via Epoch AIOriginal-source terms apply
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.