- input
- $1.25
- cached
- $0.15
- output
- $4.25
- Context
- 1Mtokens
- Released
- 2mo 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 #11
- 1501
Claude Opus 4.7Nearby rank - 1497
Gemini 3.8 FlashNearby rank - 1496
Muse Spark 1.3 - 1493Muse Spark 1.2Predecessor
- 1491Muse Spark 1.1Nearby rank
- 1447
GPT-5.4 MiniSimilar price - 1443
InklingSimilar price - 1434
GLM 5V TurboSimilar price - 1414Claude Haiku 4.5Similar price
Level with Muse Spark 1.2 (predecessor, overlapping 95% CIs) · Ahead of 4 of 4 similarly priced models
Arena WebDev
Best Arena rating per model · Higher is better
Arena #20
- 1679GPT-6.1 TerraSimilar price
- 1664
DeepSeek V4.1 Flash 1017Nearby rank - 1658
Kimi K3Nearby rank - 1655Muse Spark 1.3
- 1649Muse WonderNearby rank
- 1646Inkling 2 SmallNearby rank
- 1534Muse Spark 1.2Predecessor
- 1515
Qwen3.7 MaxSimilar price - 1399GLM 5V TurboSimilar price
- 1330Claude Haiku 4.5Similar price
+122 vs Muse Spark 1.2 (predecessor) · Ahead of 3 of 4 similarly priced models
Arena Vision
Best Arena rating per model · Higher is better
Arena #13
- 1294Muse Spark 1.2Predecessor
- 1293Gemini 3.8 FlashSimilar price
- 1293GPT-6.1 SolNearby rank
- 1291Muse Spark 1.3
- 1289Gemini 3 Pro PreviewNearby rank
- 1289Claude Fable 5.1Nearby rank
- 1250GPT-5.4 MiniSimilar price
- 1242
Grok 4.3Similar price - 1232GLM 5V TurboSimilar price
Level with Muse Spark 1.2 (predecessor, overlapping 95% CIs) · 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 #16
- 1480Muse Spark 1.1Earlier version
- 1478Claude Sonnet 4.6Nearby rank
- 1474GPT-5.6 TerraNearby rank
- 1474Muse Spark 1.3
- 1474GPT-6 AstraNearby rank
- 1471GPT-5.4Nearby rank
- 1469Gemini 3.8 FlashSimilar price
- 1450Kimi K2.6Similar price
- 1419Claude Haiku 4.5Similar price
- 1406GLM 5V TurboSimilar price
Level with Muse Spark 1.1 (earlier version, overlapping 95% CIs) · Ahead of 3 of 4 similarly priced models
Agent Arena
Best IPS score per model · Higher is better
Arena #21
- +0.120GPT-6.1 TerraSimilar price
- +0.024
MiMo-V2.6-ProNearby rank - +0.023Gemini 3.8 FlashNearby rank
- +0.022Muse Spark 1.3
- +0.020DeepSeek V4.1 Flash 1017Nearby rank
- +0.020Claude Sonnet 5Nearby rank
- +0.002GLM 5.2Similar price
- -0.072Qwen3.7 MaxSimilar price
- -0.082Muse Spark 1.2Predecessor
- -0.173InklingSimilar price
+0.104 vs Muse Spark 1.2 (predecessor) · Ahead of 3 of 4 similarly priced models
FrontierMath-Tier-4-v2-Private
Best score per model · Higher is better
#23 of 72
- 92.7%GPT-6.1 TerraSimilar price
- 46.3%Claude Haiku 5.5Nearby rank
- 46.3%DeepSeek V4.1 Flash 1017Nearby rank
- 46.3%Muse Spark 1.3
- 46.0%GPT-5.2 ProNearby rank
- 39.0%Kimi K3Nearby rank
- 34.1%Qwen3.7 MaxSimilar price
- 29.3%GLM 5.2Similar price
- 4.9%InklingSimilar price
Ahead of 3 of 4 similarly priced models
FrontierMath-Tiers-1-3-v2-Private
Best score per model · Higher is better
#24 of 87
- 90.9%GPT-6.1 TerraSimilar price
- 75.1%Claude Haiku 5.5Nearby rank
- 74.7%
Qwen3.8 MaxNearby rank - 74.4%Muse Spark 1.3
- 74.0%GPT-5.2 ProNearby rank
- 72.2%Kimi K3Nearby rank
- 64.6%Qwen3.7 MaxSimilar price
- 36.8%GLM 5.1Similar price
- 33.3%InklingSimilar price
Ahead of 3 of 4 similarly priced models
OTIS Mock AIME 2024-2025
Best score per model · Higher is better
#17 of 158
- 100.0%GPT-6.1 TerraSimilar price
- 99.4%Qwen3.8 MaxNearby rank
- 99.4%DeepSeek V4.1 Flash 1017Nearby rank
- 99.2%Muse Spark 1.3
- 99.2%Grok 4.6Nearby rank
- 98.9%Claude Haiku 5.5Nearby rank
- 95.6%Qwen3.7 MaxSimilar price
- 93.3%GLM 5.1Similar price
- 66.7%Claude Haiku 4.5Similar price
Ahead of 3 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
Arena WebDev
Score by reasoning effort · Higher is better
Arena Vision
Score by reasoning effort · Higher is better
FrontierMath-Tier-4-v2-Private
Score by reasoning effort · Higher is better
FrontierMath-Tiers-1-3-v2-Private
Score by reasoning effort · Higher is better
All published scores
- Arena TextArena #111496
Level with Muse Spark 1.2 (predecessor, overlapping 95% CIs)
- max149695% CI 1491 to 1501 · 26,111 votes
- Arena WebDevArena #201655
+122 vs Muse Spark 1.2 (predecessor)
- xhigh162095% CI 1613 to 1627 · 11,653 votes
- max165595% CI 1648 to 1662 · 12,122 votes
- Arena VisionArena #131291
Level with Muse Spark 1.2 (predecessor, overlapping 95% CIs)
- xhigh128895% CI 1280 to 1296 · 7,522 votes
- max129195% CI 1283 to 1300 · 7,683 votes
- Arena DocumentArena #161474
Level with Muse Spark 1.1 (earlier version, overlapping 95% CIs)
- max147495% CI 1466 to 1482 · 7,205 votes
- Agent ArenaArena #21+0.022
+0.104 vs Muse Spark 1.2 (predecessor)
- max+0.02295% CI +0.013 to +0.030 · 8,894,242 observations
- FrontierMath-Tier-4-v2-Private#23 of 7246.3%
- xhigh41.5%
- max46.3%
Source:Epoch AI evaluation - FrontierMath-Tiers-1-3-v2-Private#24 of 8774.4%
- xhigh74.4%
- max74.0%
Source:Epoch AI evaluation - OTIS Mock AIME 2024-2025#17 of 15899.2%
- xhigh99.2%
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.