Models › Coding agents
Coding agents
For picking a coding model: SWE-bench Verified and Aider scores side by side.
- Fastest first token112msDeepSeek-V4.1-FlashTogether AI via huggingface.co, 2026-10-25
- Cheapest with tool calls$0.010per 1M inQwen3-Embedding-8BDeepInfra via huggingface.co, 2026-10-25
- Most fine-tuned 2025 base2.6kfine-tunesQwen3-4B-Instruct-2507Hugging Face base-model tags, 2026-10-25
Models with a SWE-bench Verified or Aider Polyglot row (third-party board or the author's card). Sorted by SWE-bench Verified.
The floor
Every view leaves out test fixtures, models whose size didn't parse, and closed models (unless you ask for them). $0 endpoints show as a free tier and never count as cheapest; classifiers and guards never enter a speed sort. How the rank works.
Kind: All kinds · LLMs · Embeddings · Decision models · Closed
42 models match. Data observed 2026-10-25 06:54 UTC, Hub snapshot 2026-10-25. Hover a figure for its source and date.
| # | Evidence | ||||||
|---|---|---|---|---|---|---|---|
| 1 | Kimi-K2.6moonshotai | 80.2 | — | $0.45 | 18 | ~616 GB | |
| 2 | Inkling-Smallthinkingmachines | 80.2 | — | $0.45 | 2 | ~160 GB | |
| 3 | Inklingthinkingmachines | 77.6 | — | $0.95 | 4 | ~571 GB | |
| 4 | Qwen3.6-27BQwen | 77.2 | — | $0.20 | 9 | ~17 GB | |
| 5 | Muse-Glimmer-30Bmeta-models | 76 | — | $0.30 | 3 | ~18 GB | |
| 6 | MiniMax-M2.5MiniMaxAI | 75.8 | — | $0.27 | 8 | ~137 GB | |
| 7 | Step-3.5-Flashstepfun-ai | 74.4 | — | $0.10 | 2 | ~120 GB | |
| 8 | MiniMax-M2.1MiniMaxAI | 74 | — | $0.30 | 2 | ~137 GB | |
| 9 | GLM-4.7zai-org | 73.8 | — | $0.32 | 7 | ~215 GB | |
| 10 | Qwen3.6-35B-A3BQwen | 73.4 | — | $0.050 | 12 | ~22 GB | |
| 11 | GLM-5zai-org | 72.8 | — | $0.60 | 10 | ~452 GB | |
| 12 | Kimi-K2.5moonshotai | 70.8 | — | $0.45 | 7 | ~616 GB | |
| 13 | Ornith-1.5-9Bornith-ai | 70.6 | — | — | 1 | ~5.8 GB | |
| 14 | DeepSeek-V3.2deepseek-ai | 70 | — | $0.21 | 11 | ~411 GB | |
| 15 | Qwen3-Coder-480B-A35B-InstructQwen | 69.6 | — | $0.22 | 4 | ~288 GB | |
| 16 | Kimi-K2-Instruct-0905moonshotai | 69.2 | — | $0.60 | 1 | ~616 GB | |
| 17 | GLM-4.6zai-org | 68.2 | — | $0.40 | 5 | ~214 GB | |
| 18 | DeepSeek-V3.2-Expdeepseek-ai | 67.8 | 74.5 | $0.27 | 1 | ~411 GB | |
| 19 | DeepSeek-V3.1-Basedeepseek-ai | 66 | 76.3 | — | 0 | ~411 GB | |
| 20 | GLM-4.5zai-org | 64.2 | — | $0.60 | 1 | ~215 GB | |
| 21 | Nanbeige4.2-3BNanbeige | 63.6 | — | — | 0 | ~2.5 GB | |
| 22 | Kimi-K2-Thinkingmoonshotai | 63.4 | — | $0.60 | 3 | ~616 GB | |
| 23 | MiniMax-M2MiniMaxAI | 61 | — | $0.26 | 3 | ~137 GB | |
| 24 | Qwen3-Coder-30B-A3B-InstructQwen | 60.4 | — | $0.070 | 3 | ~18 GB | |
| 25 | GLM-4.7-Flashzai-org | 59.2 | — | $0.060 | 5 | ~19 GB | |
| 26 | DeepSeek-R1-0528deepseek-ai | 57.6 | 71.4 | $0.50 | 5 | ~411 GB | |
| 27 | Kimi-K2-Instructmoonshotai | 53.4 | 59.1 | $0.57 | 2 | ~616 GB | |
| 28 | K-EXAONE-236B-A23BLGAI-EXAONE | 49.4 | — | — | 0 | ~142 GB | |
| 29 | DeepSeek-R1deepseek-ai | 49.2 | 56.9 | $0.70 | 1 | ~411 GB | |
| 30 | Qwen2.5-Coder-32B-InstructQwen | 47 | 16.4 | $0.060 | 3 | ~20 GB | |
| 31 | DeepSeek-V3-0324deepseek-ai | 42 | 55.1 | $0.24 | 4 | ~411 GB | |
| 32 | gpt-oss-120bopenai | 26 | 41.8 | $0.030 | 27 | ~70 GB | |
| 33 | DeepSeek-V2.5deepseek-ai | 16.8 | 72.2 | — | 0 | ~141 GB | |
| 34 | Qwen3-235B-A22BQwen | — | 59.6 | $0.20 | 2 | ~141 GB | |
| 35 | Qwen3-235B-A22B-Instruct-2507Qwen | — | 57.3 | $0.087 | 7 | ~141 GB | |
| 36 | Qwen3-Next-80B-A3B-InstructQwen | — | 49.8 | $0.072 | 4 | ~49 GB | |
| 37 | Qwen3-32BQwen | — | 40 | $0.080 | 5 | ~20 GB | |
| 38 | Qwen3-30B-A3B-Instruct-2507Qwen | — | 35.6 | $0.090 | 4 | ~18 GB | |
| 39 | QwQ-32BQwen | — | 26.2 | — | 1 | ~20 GB | |
| 40 | Llama-4-Maverick-17B-128E-Instructmeta-llama | — | 15.6 | $0.19 | 3 | ~241 GB | |
| 41 | Qwen3-4B-Instruct-2507Qwen | — | 12.9 | $0.010 | 1 | ~2.4 GB | |
| 42 | gemma-3-27b-itgoogle | — | 4.9 | $0.080 | 6 | ~17 GB |
Evidence marks, left to right: coding, tools, reasoning, long docs, vision, chat. measured on a third-party board, author-reported, structural, none yet. Each mark links to its job from the model page. INT4 is an estimate of the weights alone (method).