- Context
- 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
#18 of 28
- 74.1%
- 73.8%
- 73.6%
- 72.7%
- 69.9%
- 69.6%
- 69.0%
- 68.5%
- 67.5%
- 67.2%
- 53.8%Claude Sonnet 5
Arena Text
Best Arena rating per model · Higher is better
Arena #60
- 1526Gemini 4 CarbonSimilar price
- 1487Kimi K3Similar price
- 1482GPT-6.1 SolSimilar price
- 1472Claude Sonnet 4.6Predecessor
- 1465GLM 5.1Nearby rank
- 1464
DeepSeek V4 Pro 0813Nearby rank - 1461Claude Sonnet 5
- 1461Kimi K2.6Nearby rank
- 1460
Qwen3.6 Max PreviewNearby rank - 1354
Command ASimilar price
-11 vs Claude Sonnet 4.6 (predecessor) · Ahead of 1 of 4 similarly priced models
Arena WebDev
Best Arena rating per model · Higher is better
Arena #54
- 1768Gemini 4 CarbonSimilar price
- 1743GPT-6.1 SolSimilar price
- 1676
Qwen3.8 Max (0902)Similar price - 1658Kimi K3Similar price
- 1547Claude Opus 4.6Nearby rank
- 1541
Muse Spark 1.1Nearby rank - 1539Claude Sonnet 5
- 1538Gemini 3.6 FlashNearby rank
- 1534Muse Spark 1.2Nearby rank
- 1522Claude Sonnet 4.6Predecessor
+17 vs Claude Sonnet 4.6 (predecessor) · Ahead of 0 of 4 similarly priced models
Arena Vision
Best Arena rating per model · Higher is better
Arena #45
- 1293GPT-6.1 SolSimilar price
- 1280Grok 4.5Similar price
- 1280Gemini 3.1 Pro PreviewSimilar price
- 1275Claude Sonnet 4.6Predecessor
- 1263Claude Sonnet 5.5Nearby rank
- 1262Qwen3.7 PlusNearby rank
- 1261Claude Sonnet 5
- 1258GPT-6 SolNearby rank
- 1256
Seed 2.0 ProNearby rank - 1199
Mistral Medium 3.5Similar price
-14 vs Claude Sonnet 4.6 (predecessor) · 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 #20
- 1480Muse Spark 1.1Similar price
- 1478Claude Sonnet 4.6Predecessor
- 1474GPT-5.6 TerraSimilar price
- 1471GPT-5.4Nearby rank
- 1469Gemini 3.8 FlashNearby rank
- 1469Claude Sonnet 5
- 1468Gemini 3.7 FlashNearby rank
- 1467Claude Opus 4.5Nearby rank
- 1466Grok 4.5Similar price
- 1459Gemini 3.1 Pro PreviewSimilar price
Level with Claude Sonnet 4.6 (predecessor, overlapping 95% CIs) · Ahead of 0 of 4 similarly priced models
Agent Arena
Best IPS score per model · Higher is better
Arena #23
- +0.170Gemini 4 CarbonSimilar price
- +0.125GPT-6.1 SolSimilar price
- +0.026Grok 4.7Similar price
- +0.022Muse Spark 1.3Nearby rank
- +0.020DeepSeek V4.1 Flash 1017Nearby rank
- +0.020Claude Sonnet 5
- +0.018Muse WonderNearby rank
- +0.017Muse EmberNearby rank
- +0.006Kimi K3Similar price
Ahead of 0 of 4 similarly priced models
FrontierMath-Tier-4-v2-Private
Best score per model · Higher is better
#35 of 72
- 100.0%GPT-6.1 SolSimilar price
- 95.1%Gemini 4 CarbonSimilar price
- 39.0%Kimi K3Similar price
- 34.1%Qwen3.8 Max (0902)Similar price
- 31.7%Grok 4.6Nearby rank
- 31.7%GPT-5.2Nearby rank
- 29.3%Claude Sonnet 5
- 29.3%GLM 5.2Nearby rank
- 29.3%GLM 5.3Nearby rank
- 2.4%Claude Sonnet 4.5Earlier version
+26.8 pts vs Claude Sonnet 4.5 (earlier version) · Ahead of 0 of 4 similarly priced models
FrontierMath-Tiers-1-3-v2-Private
Best score per model · Higher is better
#36 of 87
- 93.7%GPT-6.1 SolSimilar price
- 90.5%Gemini 4 CarbonSimilar price
- 72.2%Kimi K3Similar price
- 66.0%Claude Opus 4.6Nearby rank
- 66.0%Grok 4.6Similar price
- 65.6%Claude Sonnet 5
- 65.6%Qwen3.8 Max (0902)Nearby rank
- 64.6%DeepSeek V4 Pro 0813Nearby rank
- 23.9%Claude Sonnet 4.5Earlier version
+41.8 pts vs Claude Sonnet 4.5 (earlier version) · Ahead of 0 of 4 similarly priced models
GPQA diamond
Best score per model · Higher is better
#40 of 161
- 96.5%Gemini 4 CarbonSimilar price
- 95.4%GPT-6.1 SolSimilar price
- 94.0%Grok 4.6Similar price
- 93.1%Kimi K3Similar price
- 90.6%Claude Opus 5.5Nearby rank
- 90.5%Claude Opus 4.6Nearby rank
- 90.5%Claude Sonnet 5
- 90.5%GPT-6 LunaNearby rank
- 90.2%Claude Opus 4.7Nearby rank
- 87.4%Claude Sonnet 4.6Predecessor
+3.2 pts vs Claude Sonnet 4.6 (predecessor) · Ahead of 0 of 4 similarly priced models
OTIS Mock AIME 2024-2025
Best score per model · Higher is better
#42 of 158
- 100.0%Gemini 4 ArgonSimilar price
- 100.0%GPT-6 SolSimilar price
- 100.0%Qwen3.8 Max (0902)Similar price
- 97.2%Kimi K3Similar price
- 95.6%Kimi K2.7 CodeNearby rank
- 95.6%Qwen3.7 MaxNearby rank
- 94.7%Claude Sonnet 5
- 94.4%Claude Opus 4.6Nearby rank
- 94.4%DeepSeek V4 Flash 0731Nearby rank
- 85.8%Claude Sonnet 4.6Predecessor
+8.9 pts vs Claude Sonnet 4.6 (predecessor) · Ahead of 0 of 4 similarly priced models
SimpleQA Verified
Best score per model · Higher is better
#61 of 87
- 77.7%Gemini 4 CarbonSimilar price
- 73.9%GPT-6.1 SolSimilar price
- 56.0%Grok 4.7Similar price
- 50.6%Kimi K3Similar price
- 35.5%Claude Sonnet 4.6Predecessor
- 34.2%GLM 5.2Nearby rank
- 34.0%GLM 5.1Nearby rank
- 33.7%Claude Sonnet 5
- 33.6%DeepSeek V4 Flash 0731Nearby rank
- 33.2%Grok 4.3Nearby rank
-1.8 pts vs Claude Sonnet 4.6 (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
- 78.8%
- Average cost / attempt
- $26.40
- 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 | 53.8% | 78.8% | $26.40 |
| mini-swe-agent · xhigh | 49.7% | 75.2% | $11.89 |
| mini-swe-agent · high | 48.2% | 79.6% | $7.43 |
| mini-swe-agent · medium | 39.8% | 64.6% | $4.08 |
| mini-swe-agent · low | 30.5% | 57.5% | $2.19 |
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
GPQA diamond
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 #601461
-11 vs Claude Sonnet 4.6 (predecessor)
- high146195% CI 1457 to 1465 · 53,013 votes
- Arena WebDevArena #541539
+17 vs Claude Sonnet 4.6 (predecessor)
- high153995% CI 1533 to 1545 · 16,790 votes
- Arena VisionArena #451261
-14 vs Claude Sonnet 4.6 (predecessor)
- high126195% CI 1254 to 1268 · 14,778 votes
- Arena DocumentArena #201469
Level with Claude Sonnet 4.6 (predecessor, overlapping 95% CIs)
- high146995% CI 1462 to 1475 · 11,044 votes
- Agent ArenaArena #23+0.020
- high+0.02095% CI +0.008 to +0.032 · 5,237,154 observations
- FrontierMath-Tier-4-v2-Private#35 of 7229.3%
+26.8 pts vs Claude Sonnet 4.5 (earlier version)
- max29.3%
Source:Epoch AI evaluation - FrontierMath-Tiers-1-3-v2-Private#36 of 8765.6%
+41.8 pts vs Claude Sonnet 4.5 (earlier version)
- max65.6%
Source:Epoch AI evaluation - GPQA diamond#40 of 16190.5%
+3.2 pts vs Claude Sonnet 4.6 (predecessor)
- xhigh90.5%
- max80.3%
Source:Epoch AI evaluation - OTIS Mock AIME 2024-2025#42 of 15894.7%
+8.9 pts vs Claude Sonnet 4.6 (predecessor)
- xhigh94.7%
- max80.0%
Source:Epoch AI evaluation - SimpleQA Verified#61 of 8733.7%
-1.8 pts vs Claude Sonnet 4.6 (predecessor)
- xhigh32.9%
- max33.7%
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.