AI Coding Model Leaderboard · ranked at each model's recommended effort
AI coding model leaderboard: every model compared at its own best-value ★ reasoning effort, not just its top setting. Ability, cost and task time in one view, plus the coding plans that cover them.
Leaderboard: every model at its ★ recommended effort
★ is the best-value effort among a model's own efforts, not its top effort; the top effort is listed separately. Ability is the ModelPlan ability score: a frozen scale where 50 is the September 2026 frontier average and 10 points are about one standard deviation; not a percentage, so it can fall below 0 or rise above 100. Cost is the API cost of one reference coding task in USD; verdicts compare each maker's lineup. FAQ
| # | Model | Maker | Recommended effort | Ability | Cost / task | Top effort (ability · cost) | Verdict |
|---|---|---|---|---|---|---|---|
| 1 | Gemini 4 Argon | ★ High | 67.5 | $4.28 | same as ★ | Strongest · Best value | |
| 2 | GPT-6 Astra | OpenAI | ★ Med | 67.4 | $2.88 | Max · 73.7 · $7.08 | Strongest |
| 3 | GPT-6.1 Sol | OpenAI | ★ Med | 65.8 | $0.533 | Max · 71.8 · $1.40 | |
| 4 | Kimi K3.1 Pro | Kimi | ★ Max | 65.1 | $4.24 | same as ★ | |
| 5 | Claude Opus 5.5 | Anthropic | ★ Med | 64.9 | $2.50 | Max · 77.3 · $10.82 | Strongest · Best value |
| 6 | Claude Fable 5.1 | Anthropic | ★ Med | 62.6 | $5.06 | Max · 70.4 · $12.84 | Outclassed |
| 7 | Muse Wonder | Meta | ★ Max | 62.1 | $6.66 | same as ★ | |
| 8 | Claude Sonnet 5.5 | Anthropic | ★ High | 58.3 | $1.39 | Max · 74.9 · $11.20 | |
| 9 | DeepSeek V4.1 Flash 1017 | DeepSeek | ★ Max | 55.3 | $0.526 | same as ★ | |
| 10 | GPT-6.1 Terra | OpenAI | ★ Med | 52.6 | $0.147 | Max · 70.3 · $1.07 | Best value |
| 11 | Grok 4.7 | SpaceXAI | ★ High | 52.4 | $5.22 | XHigh · 54.2 · $7.16 | |
| 12 | Claude Haiku 5.5 | Anthropic | ★ XHigh | 50.6 | $0.459 | Max · 50.9 · $1.37 | Cheapest |
| 13 | Qwen3.8 Max (0902) | Alibaba | ★ Max | 49.4 | $11.53 | same as ★ | |
| 14 | GLM-5.3 | 智谱 AI | ★ Max | 47.4 | $4.55 | same as ★ | Strongest · Best value |
| 15 | MiMo-V2.6-Pro | Xiaomi | ★ Max | 44.7 | $0.296 | same as ★ | Strongest · Best value |
| 16 | Gemini 3.8 Flash | ★ High | 42.8 | $3.47 | same as ★ | Cheapest | |
| 17 | GLM-5.3-Flash | 智谱 AI | ★ Max | 42.0 | $0.457 | same as ★ | Cheapest |
| 18 | MiMo-V2.6-Flash | Xiaomi | ★ Max | 40.8 | $0.148 | same as ★ | Cheapest |
| 19 | GPT-6 Luna | OpenAI | ★ High | 35.4 | $0.084 | Max · 43.6 · $0.173 | Cheapest |
| 20 | MiniMax-M3 | MiniMax | ★ Max | 31.2 | $1.48 | same as ★ |
Which model is worth it, per maker
- OpenAI: GPT-6 Luna → GPT-6.1 Terra: ability +17.3, cost ×1.75; GPT-6.1 Terra → GPT-6.1 Sol: ability +13.2, cost ×3.62; GPT-6.1 Sol → GPT-6 Astra: ability +1.6, cost ×5.41
- Anthropic: Claude Haiku 5.5 → Claude Sonnet 5.5: ability +7.7, cost ×3.02; Claude Sonnet 5.5 → Claude Opus 5.5: ability +6.6, cost ×1.81. Skip: Claude Fable 5.1
- 智谱 AI: GLM-5.3-Flash → GLM-5.3: ability +5.5, cost ×9.94
- Google: Gemini 3.8 Flash → Gemini 4 Argon: ability +24.7, cost ×1.23
- Xiaomi: MiMo-V2.6-Flash → MiMo-V2.6-Pro: ability +3.9, cost ×2.00
Data published 2026-11-14 09:48 UTC · All data is also available through the public API