- input
- $0.14
- cached
- $0.05
- output
- $1.00
- Context
- 262Ktokens
- Released
- 8mo ago
- Size
- 35Bparameters
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 #175
- 1483
GLM 5.3Similar price - 1425
DeepSeek V3.2 ExpSimilar price - 1415
MiniMax M2.7Similar price - 1395DeepSeek V3 0324Nearby rank
- 1395
Qwen3 VL 235B A22B ThinkingNearby rank - 1394
Qwen3.5-35B-A3B - 1393
Step 3.5 FlashNearby rank - 1392
MiMo-V2-FlashNearby rank - 1347Qwen3 32BPredecessor
- 1303
Mistral Small 3.1 24BSimilar price
+47 vs Qwen3 32B (predecessor) · Ahead of 1 of 4 similarly priced models
Arena WebDev
Best Arena rating per model · Higher is better
Arena #143
- 1622GLM 5.3Similar price
- 1398MiniMax M2.7Similar price
- 1358
Hy3 previewSimilar price - 1270DeepSeek V3.2 ExpSimilar price
- 1263Mistral Medium 3.5Nearby rank
- 1256
Gemini 3.1 Flash Lite PreviewNearby rank - 1253Qwen3.5-35B-A3B
- 1245
GPT-5.1-Codex-MiniNearby rank - 1244Qwen3.5-FlashNearby rank
Ahead of 0 of 4 similarly priced models
Arena Vision
Best Arena rating per model · Higher is better
Not ranked on Arena Vision yet
Arena Search
Best Arena rating per model · Higher is better
Not ranked on Arena Search yet
Agent Arena
Best IPS score per model · Higher is better
Not ranked on Agent Arena yet
GPQA diamond
Best score per model · Higher is better
#75 of 161
- 91.6%GPT-5.6 LunaSimilar price
- 90.9%GLM 5.3Similar price
- 84.8%Qwen3.6 35B A3BNearby rank
- 84.2%
Kimi K2 Thinking TurboNearby rank - 83.5%Qwen3.5-35B-A3B
- 83.4%DeepSeek ReasonerNearby rank
- 83.3%Gemini 3.5 Flash LiteNearby rank
- 75.8%Gemma 4 31BSimilar price
- 71.2%DeepSeek V3Similar price
- 65.7%Qwen3 32BPredecessor
+17.7 pts vs Qwen3 32B (predecessor) · Ahead of 2 of 4 similarly priced models
OTIS Mock AIME 2024-2025
Best score per model · Higher is better
#96 of 158
- 98.3%GPT-5.6 LunaSimilar price
- 91.1%GLM 5.3Similar price
- 73.3%Gemma 4 31BSimilar price
- 71.1%MiniMax M3Nearby rank
- 70.8%Gemini 2.5 Flash Preview 05-20Nearby rank
- 70.0%Qwen3.5-35B-A3B
- 69.4%
Seed 1.6Nearby rank - 68.9%
Claude Opus 4.1Nearby rank - 66.9%Qwen3 32BPredecessor
- 48.9%DeepSeek V3Similar price
+3.1 pts vs Qwen3 32B (predecessor) · Ahead of 1 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 #1751394
+47 vs Qwen3 32B (predecessor)
- default139495% CI 1390 to 1398 · 30,206 votes
- Arena WebDevArena #1431253
- default125395% CI 1235 to 1270 · 1,612 votes
- GPQA diamond#75 of 16183.5%
+17.7 pts vs Qwen3 32B (predecessor)
- none81.2%
- default83.5%
Source:Epoch AI evaluation - OTIS Mock AIME 2024-2025#96 of 15870.0%
+3.1 pts vs Qwen3 32B (predecessor)
- none70.0%
- default54.4%
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.