- Context
- 1.1Mtokens
- Released
- 4mo 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
#4 of 28
- 74.1%
- 73.8%
- 73.6%
- 72.7%GPT-5.6 Sol
- 69.9%
- 69.6%
- 69.0%
- 68.5%
- 67.5%
- 67.2%
Arena Text
Best Arena rating per model · Higher is better
Arena #22
- 1511
Muse WonderSimilar price - 1506Claude Opus 5.5Similar price
- 1503Claude Opus 4.6Similar price
- 1487Gemini 3.1 Pro PreviewSimilar price
- 1486Gemini 3 Pro PreviewNearby rank
- 1485GPT-5.6 Sol
- 1484
Qwen3.8 MaxNearby rank - 1483Kimi K3.1 ProNearby rank
- 1481GPT-5.5Predecessor
Level with GPT-5.5 (predecessor, overlapping 95% CIs) · Ahead of 0 of 4 similarly priced models
Arena WebDev
Best Arena rating per model · Higher is better
Arena #33
- 1814Claude Opus 5.5Similar price
- 1691Claude Opus 5Similar price
- 1649Muse WonderSimilar price
- 1621
DeepSeek V4.1 FlashNearby rank - 1618Grok 4.6Nearby rank
- 1615GPT-5.6 Sol
- 1609GLM 5.3 FlashNearby rank
- 1609Muse EmberNearby rank
- 1512GPT-5.5Predecessor
- 1446Gemini 3.1 Pro PreviewSimilar price
+103 vs GPT-5.5 (predecessor) · Ahead of 1 of 4 similarly priced models
Arena Vision
Best Arena rating per model · Higher is better
Arena #16
- 1300Claude Opus 4.7Similar price
- 1299Claude Opus 4.6Similar price
- 1295Claude Opus 5.5Similar price
- 1289Gemini 3 Pro PreviewNearby rank
- 1289Claude Fable 5.1Nearby rank
- 1288GPT-5.6 Sol
- 1286Claude Opus 4.8Nearby rank
- 1285Claude Opus 5Nearby rank
- 1285GPT-5.5Predecessor
- 1280Gemini 3.1 Pro PreviewSimilar price
Level with GPT-5.5 (predecessor, overlapping 95% CIs) · Ahead of 0 of 4 similarly priced models
Arena Search
Best Arena rating per model · Higher is better
Arena #1
- 1255GPT-5.6 Sol
- 1248GPT-6.1 SolNearby rank
- 1245GPT-6 AstraNearby rank
- 1213Claude Opus 5Similar price
- 1213Claude Opus 5.5Similar price
- 1210Claude Opus 4.7Similar price
- 1142
Sonar ProSimilar price
Ahead of 4 of 4 similarly priced models
Arena Document
Best Arena rating per model · Higher is better
Arena #7
- 1497Claude Opus 4.6Similar price
- 1494Claude Opus 4.7Similar price
- 1486GPT-5.6 Sol
- 1486Claude Opus 5Similar price
- 1485Claude Opus 4.8Nearby rank
- 1481GPT-5.5Predecessor
- 1459Gemini 3.1 Pro PreviewSimilar price
Level with GPT-5.5 (predecessor, overlapping 95% CIs) · Ahead of 1 of 4 similarly priced models
Agent Arena
Best IPS score per model · Higher is better
Arena #16
- +0.195Claude Opus 5.5Similar price
- +0.072Claude Opus 5Similar price
- +0.053
Hy4Nearby rank - +0.049Kimi K3.1 ProNearby rank
- +0.046GPT-5.6 Sol
- +0.026Grok 4.7Nearby rank
- +0.025
Inkling 2 SmallNearby rank - +0.018Muse WonderSimilar price
- -0.045Gemini 3.1 Pro PreviewSimilar price
Ahead of 2 of 4 similarly priced models
ARC-AGI-2
Best score per model · Higher is better
#6 of 85
- 93.8%Claude Fable 5.2Nearby rank
- 93.3%Claude Opus 5.5Similar price
- 92.5%GPT-5.6 Sol
- 91.3%Gemini 4 ArgonNearby rank
- 90.0%Claude Fable 5.1Nearby rank
- 88.3%Claude Opus 5Similar price
- 85.0%GPT-5.5Predecessor
- 77.1%Gemini 3.1 Pro PreviewSimilar price
- 75.8%Claude Opus 4.7Similar price
+7.5 pts vs GPT-5.5 (predecessor) · Ahead of 3 of 4 similarly priced models
FrontierMath-Tier-4-v2-Private
Best score per model · Higher is better
#10 of 72
- 95.0%Claude Opus 5.5Similar price
- 90.0%GPT-6 SolNearby rank
- 87.8%Claude Fable 5.1Nearby rank
- 82.9%GPT-5.6 Sol
- 80.5%Claude Sonnet 5.5Nearby rank
- 78.0%Gemini 4 ArgonNearby rank
- 73.2%Claude Opus 5Similar price
- 72.5%GPT-5.5Predecessor
- 56.1%Claude Opus 4.8Similar price
- 26.8%Gemini 3.1 Pro PreviewSimilar price
+10.4 pts vs GPT-5.5 (predecessor) · Ahead of 3 of 4 similarly priced models
FrontierMath-Tiers-1-3-v2-Private
Best score per model · Higher is better
#9 of 87
- 91.2%Claude Opus 5.5Similar price
- 90.2%Claude Fable 5.1Nearby rank
- 89.8%GPT-6 SolNearby rank
- 89.1%GPT-5.6 Sol
- 88.8%Claude Sonnet 5.5Nearby rank
- 87.7%GPT-5.5 ProNearby rank
- 85.6%Claude Opus 5Similar price
- 85.3%GPT-5.5Predecessor
- 80.0%Claude Opus 4.8Similar price
- 59.6%Gemini 3.1 Pro PreviewSimilar price
+3.9 pts vs GPT-5.5 (predecessor) · Ahead of 3 of 4 similarly priced models
GPQA diamond
Best score per model · Higher is better
#16 of 161
- 94.4%Gemini 3.1 Pro PreviewSimilar price
- 93.9%Claude Fable 5.2Nearby rank
- 93.9%Claude Opus 5Similar price
- 93.5%GPT-5.6 Sol
- 93.4%Grok 4.5Nearby rank
- 93.3%GPT-5.6 TerraNearby rank
- 91.0%Claude Opus 4.8Similar price
- 90.7%GPT-5.5Predecessor
- 90.6%Claude Opus 5.5Similar price
+2.8 pts vs GPT-5.5 (predecessor) · Ahead of 2 of 4 similarly priced models
OSWorld 2.0
Best score per model · Higher is better
#2 of 9
- 31.4%Claude Opus 5Similar price
- 27.3%GPT-5.6 Sol
- 18.5%Claude Opus 4.8Similar price
- 13.9%Claude Opus 4.7Similar price
- 13.0%GPT-5.5Predecessor
- 9.3%Claude Sonnet 4.6Similar price
+14.3 pts vs GPT-5.5 (predecessor) · Ahead of 3 of 4 similarly priced models
OTIS Mock AIME 2024-2025
Best score per model · Higher is better
#1 of 158
- 100.0%Claude Opus 5.5Similar price
- 100.0%Gemini 4 ArgonNearby rank
- 100.0%Gemini 4 CarbonNearby rank
- 100.0%GPT-5.6 Sol
- 100.0%GPT-6 AstraNearby rank
- 100.0%GPT-6 SolNearby rank
- 98.9%Claude Opus 5Similar price
- 98.3%Claude Opus 4.8Similar price
- 95.6%Gemini 3.1 Pro PreviewSimilar price
- 84.4%GPT-5.5Predecessor
+15.6 pts vs GPT-5.5 (predecessor) · Ahead of 3 of 4 similarly priced models
SimpleQA Verified
Best score per model · Higher is better
#10 of 87
- 73.5%Gemini 3.1 Pro PreviewSimilar price
- 72.2%Claude Opus 5.5Similar price
- 70.7%Claude Fable 5Nearby rank
- 69.7%Gemini 3.8 FlashNearby rank
- 69.7%GPT-5.6 Sol
- 69.2%Gemini 3.7 FlashNearby rank
- 66.8%Gemini 3 Flash PreviewNearby rank
- 63.0%GPT-5.5Predecessor
- 59.9%Claude Opus 5Similar price
- 53.0%Claude Opus 4.8Similar price
+6.7 pts vs GPT-5.5 (predecessor) · Ahead of 2 of 4 similarly priced models
DeepSWE 1.1
Primary coding-agent benchmark · 113 tasks · Data generated 2026-09-22
- pass@4
- 85.8%
- Average cost / attempt
- $8.39
- Benchmark runs
- 4
Score vs cost by reasoning effort
Average cost per attempt (USD) against pass rate
All published configurations
| Harness / reasoning | pass@1 | pass@4 | Cost / attempt |
|---|---|---|---|
| mini-swe-agent · max | 72.7% | 85.8% | $8.39 |
| mini-swe-agent · xhigh | 70.7% | 85.8% | $4.70 |
| mini-swe-agent · high | 69.4% | 86.7% | $3.47 |
| mini-swe-agent · medium | 61.1% | 80.5% | $1.86 |
| mini-swe-agent · low | 45.4% | 71.7% | $1.07 |
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
OTIS Mock AIME 2024-2025
Score by reasoning effort · Higher is better
All published scores
- Arena TextArena #221485
Level with GPT-5.5 (predecessor, overlapping 95% CIs)
- xhigh148595% CI 1481 to 1489 · 43,994 votes
- Arena WebDevArena #331615
+103 vs GPT-5.5 (predecessor)
- xhigh1615Codex Harness95% CI 1610 to 1621 · 18,914 votes
- Arena VisionArena #161288
Level with GPT-5.5 (predecessor, overlapping 95% CIs)
- xhigh128895% CI 1281 to 1295 · 12,342 votes
- Arena SearchArena #11255
- xhigh125595% CI 1249 to 1262 · 62,364 votes
- Arena DocumentArena #71486
Level with GPT-5.5 (predecessor, overlapping 95% CIs)
- xhigh148695% CI 1480 to 1493 · 9,649 votes
- Agent ArenaArena #16+0.046
- xhigh+0.04695% CI +0.035 to +0.057 · 6,641,589 observations
- ARC-AGI-2#6 of 8592.5%
+7.5 pts vs GPT-5.5 (predecessor)
- low42.5%
- medium67.1%
- high85.4%
- xhigh90.0%
- max92.5%
- FrontierMath-Tier-4-v2-Private#10 of 7282.9%
+10.4 pts vs GPT-5.5 (predecessor)
- max82.9%
Source:Epoch AI evaluation - FrontierMath-Tiers-1-3-v2-Private#9 of 8789.1%
+3.9 pts vs GPT-5.5 (predecessor)
- max89.1%
Source:Epoch AI evaluation - GPQA diamond#16 of 16193.5%
+2.8 pts vs GPT-5.5 (predecessor)
- none82.8%
- low89.9%
- max93.5%
Source:Epoch AI evaluation - OSWorld 2.0#2 of 927.3%
+14.3 pts vs GPT-5.5 (predecessor)
- max27.3%
- OTIS Mock AIME 2024-2025#1 of 158100.0%
+15.6 pts vs GPT-5.5 (predecessor)
- none68.9%
- low95.6%
- max100.0%
Source:Epoch AI evaluation - SimpleQA Verified#10 of 8769.7%
+6.7 pts vs GPT-5.5 (predecessor)
- max69.7%
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.