- input
- $1.25
- cached
- $0.15
- output
- $4.25
- Context
- 1Mtokens
- Released
- 7mo 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
Not evaluated in DeepSWE 1.1
Arena Text
Best Arena rating per model · Higher is better
Arena #14
- 1496
Muse Spark 1.3Nearby rank - 1493
Muse Spark 1.2Nearby rank - 1491
Muse Spark 1.1 - 1490
Claude Opus 5Nearby rank - 1489
Gemini 3.7 FlashNearby rank - 1447
GPT-5.4 MiniSimilar price - 1443
InklingSimilar price - 1434
GLM 5V TurboSimilar price - 1414
Claude Haiku 4.5Similar price
Ahead of 4 of 4 similarly priced models
Arena WebDev
Best Arena rating per model · Higher is better
Arena #53
- 1679
GPT-6.1 TerraSimilar price - 1552
Grok 4.5Nearby rank - 1547
Claude Opus 4.6Nearby rank - 1541
Muse Spark 1.1 - 1539
Claude Sonnet 5Nearby rank - 1538
Gemini 3.6 FlashNearby rank - 1515
Qwen3.7 MaxSimilar price - 1399
GLM 5V TurboSimilar price - 1330
Claude Haiku 4.5Similar price
Ahead of 3 of 4 similarly priced models
Arena Vision
Best Arena rating per model · Higher is better
Arena #29
- 1293
Gemini 3.8 FlashSimilar price - 1280
Gemini 3.1 Pro PreviewNearby rank - 1280
Gemini 3.6 FlashNearby rank - 1279
Muse Spark 1.1 - 1278
GPT-5.2 ChatNearby rank - 1277
GPT-5.5 InstantNearby rank - 1250
GPT-5.4 MiniSimilar price - 1242
Grok 4.3Similar price - 1232
GLM 5V TurboSimilar price
Ahead of 3 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 #11
- 1485
Claude Opus 4.8Nearby rank - 1481
GPT-5.5Nearby rank - 1480
Muse Spark 1.1 - 1478
Claude Sonnet 4.6Nearby rank - 1474
GPT-5.6 TerraNearby rank - 1469
Gemini 3.8 FlashSimilar price - 1450
Kimi K2.6Similar price - 1419
Claude Haiku 4.5Similar price - 1406
GLM 5V TurboSimilar price
Ahead of 3 of 4 similarly priced models
Agent Arena
Best IPS score per model · Higher is better
Arena #50
- +0.120
GPT-6.1 TerraSimilar price - +0.002
GLM 5.2Similar price - -0.072
Qwen3.7 MaxSimilar price - -0.076
Hy3Nearby rank - -0.080
Qwen3.7 PlusNearby rank - -0.080
Muse Spark 1.1 - -0.082
Muse Spark 1.2Nearby rank - -0.105MiniMax M3Nearby rank
- -0.173
InklingSimilar price
Ahead of 1 of 4 similarly priced models
SimpleQA Verified
Best score per model · Higher is better
#20 of 87
- 60.3%
Muse Spark 1.2Nearby rank - 59.9%
Claude Opus 5Nearby rank - 57.8%
Muse Spark 1.1 - 57.3%
GPT-6.1 TerraSimilar price - 56.0%
Grok 4.7Nearby rank - 55.8%
Qwen3.7 MaxSimilar price - 34.0%
GLM 5.1Similar price - 13.2%
Claude Haiku 4.5Similar price
Ahead of 4 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.
All published scores
- Arena TextArena #141491
- default149195% CI 1487 to 1495 · 45,562 votes
- Arena WebDevArena #531541
- default154195% CI 1534 to 1549 · 7,843 votes
- Arena VisionArena #291279
- default127995% CI 1272 to 1286 · 13,413 votes
- Arena DocumentArena #111480
- default148095% CI 1473 to 1487 · 8,934 votes
- Agent ArenaArena #50-0.080
- default-0.08095% CI -0.087 to -0.073 · 8,463,889 observations
- SimpleQA Verified#20 of 8757.8%
- default57.8%
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.