Aider Polyglot Leaderboard
coding40 models · Updated June 2026
On the Aider Polyglot benchmark, GPT-5 ranks #1 with a score of 88.0, while Qwen3 32B offers the best score-per-dollar at $0.28/1M output tokens. The full ranking, with cost per million tokens, is below.
💰 Best value
Qwen3 32B — score 40.0 at $0.28/1M output tokens
| # | Model | Score | Output / 1M | Context |
|---|---|---|---|---|
| 1 | GPT-5OpenAI | 88.0 | $10.00 | 400K |
| 2 | OpenAI o3-pro (2025-06-10)NanoGPT | 84.9 | $88.00 | 200K |
| 3 | Gemini 2.5 Pro Preview 06-05Google | 83.1 | $10.00 | 1M |
| 4 | o3OpenAI | 81.3 | $8.00 | 200K |
| 5 | Grok-4Poe | 79.6 | $15.00 | 256K |
| 6 | Gemini 2.5 Pro Preview 05-06Google | 76.9 | $10.00 | 1M |
| 7 | DeepSeek V3.2DeepSeek | 74.2 | $0.80 | 164K |
| 8 | OpenAI o4-mini highNanoGPT | 72.0 | $4.40 | 200K |
| 9 | Claude Opus 4Anthropic | 72.0 | $75.00 | 200K |
| 10 | DeepSeek-R1 (May 2025)DeepSeek | 71.4 | — | — |
| 11 | Claude 3.7 SonnetHelicone | 64.9 | $15.00 | 200K |
| 12 | o1OpenAI | 61.7 | $60.00 | 200K |
| 13 | Claude Sonnet 4Anthropic | 61.3 | $15.00 | 200K |
| 14 | o3 MiniOpenAI | 60.4 | $4.40 | 200K |
| 15 | Kimi K2 ThinkingMoonshotAI | 59.1 | $2.50 | 262K |
| 16 | DeepSeek-V3 (Mar 2025)DeepSeek | 55.1 | — | — |
| 17 | gemini-2.5-flash-preview-05-20Jiekou.AI | 55.1 | $3.15 | 1M |
| 18 | Grok 3Poe | 53.3 | $15.00 | 131K |
| 19 | GPT-4.1OpenAI | 52.4 | $8.00 | 1M |
| 20 | Claude 3.5 SonnetDigitalOcean | 51.6 | $15.00 | 200K |
| 21 | Grok 3 MiniPoe | 49.3 | $0.50 | 131K |
| 22 | chatgpt-4o-latest302.AI | 45.3 | $15.00 | 128K |
| 23 | GPT-4.5OpenAI | 44.9 | — | — |
| 24 | GPT OSS 120BNeon | 41.8 | $0.60 | 131K |
| 25 | Qwen3 32BQwen | 40.0 | $0.28 | 131K |
| 26 | o1-miniAzure | 32.9 | $4.40 | 128K |
| 27 | GPT-4.1 MiniOpenAI | 32.4 | $1.60 | 1M |
| 28 | Claude 3.5 HaikuHelicone | 28.0 | $4.00 | 200K |
| 29 | GPT-4o (Mar 2025)OpenAI | 23.1 | — | — |
| 30 | Gemini 2.0 FlashGoogle | 22.2 | $0.40 | 1M |
| 31 | Qwen MaxAlibaba | 21.8 | $6.40 | 33K |
| 32 | QwQ 32BAlibaba (China) | 20.9 | $0.86 | 131K |
| 33 | DeepSeek-V2.5DeepSeek | 17.8 | — | — |
| 34 | Qwen2.5 Coder 32B Instruct | 16.4 | $1.00 | 33K |
| 35 | Llama 4 MaverickMeta | 15.6 | $0.65 | 1M |
| 36 | Yi LightningNanoGPT | 12.9 | $0.20 | 12K |
| 37 | Codestral 25.01GitHub Models | 11.1 | $0.00 | 32K |
| 38 | GPT-4.1 NanoOpenAI | 8.9 | $0.40 | 1M |
| 39 | Gemma 3 27B ITCortecs | 4.9 | $0.30 | 131K |
| 40 | GPT-4o-miniOpenAI | 3.6 | $0.60 | 128K |
Frequently asked questions
What is the Aider Polyglot benchmark?
Aider Polyglot is a coding used to evaluate and rank large language models. This page lists 40 models by their Aider Polyglot score.
Which model is best on Aider Polyglot?
GPT-5 (OpenAI) currently leads the Aider Polyglot leaderboard with a score of 88.0.
What is the best value model on Aider Polyglot?
Qwen3 32B offers the best score-per-dollar on Aider Polyglot — a score of 40.0 at $0.28 per 1M output tokens.
Pricing is indicative — confirm with the provider before production use. Updated June 2026.