Leaderboard · Nov 2026
Best AI models for coding, ranked by developers
The main table covers models with at least 100 opinions. Less-discussed models are listed separately, because a high score from a few dozen people is a weaker signal.
Widely discussed models
Ranked by approval, adjusted for sample size. Model pages focus on coding use, not general chat.
| # | Model | Approval | Last 90 days | Opinions |
|---|---|---|---|---|
| 01 | DeepSeek V4 Flash | 83%±5 | 80% | 151 |
| 02 | Claude Opus 4.5 | 78%±3.8 | 90% | 322 |
| 03 | Claude Opus 5.5 | 75%±5.2 | 75% | 186 |
| 04 | GLM 5.2 | 75%±6 | 82% | 142 |
| 05 | DeepSeek V4 Pro | 75%±6.5 | 67% | 121 |
| 06 | GPT-5.4 | 72%±6.9 | – | 114 |
| 07 | GPT-5.5 | 69%±5 | 67% | 236 |
| 08 | GPT-5.6 Sol | 67%±4.9 | 66% | 247 |
| 09 | Claude Opus 4.6 | 63%±4.7 | 63% | 287 |
| 10 | Claude Fable 5 | 57%±5.9 | 48% | 188 |
| 11 | Claude Opus 4.8 | 49%±7.1 | 57% | 134 |
| 12 | Claude Opus 4.7 | 42%±6.7 | – | 145 |
| 13 | Claude Opus 5 | 33%±6.1 | 30% | 162 |
0183%0278%0375%0475%0575%0672%0769%0867%0963%1057%1149%1242%1333%
DeepSeek V4 Flash
Claude Opus 4.5
Claude Opus 5.5
GLM 5.2
DeepSeek V4 Pro
GPT-5.4
GPT-5.5
GPT-5.6 Sol
Claude Opus 4.6
Claude Fable 5
Claude Opus 4.8
Claude Opus 4.7
Claude Opus 5
Shown: models with at least 100 opinions. Approval is the positive share of opinions that took a side; rank uses the lower bound of a 90% confidence interval.
Less-discussed models
Fewer than 100 opinions each. Promising, but treat the scores as early.
| # | Model | Approval | Last 90 days | Opinions |
|---|---|---|---|---|
| 14 | GLM 5.3 | 85%±7.2 | 85% | 67 |
| 15 | Qwen 3 Coder | 87%±8.9 | – | 39 |
| 16 | Gemini 3 Flash | 85%±10.5 | – | 31 |
| 17 | DeepSeek V4.1 Flash | 80%±8.5 | 80% | 60 |
| 18 | GLM 5.1 | 78%±8.3 | – | 67 |
| 19 | Qwen 3.6 | 76%±7.2 | 71% | 94 |
| 20 | Qwen 3.6 27B | 76%±8.1 | 56% | 76 |
| 21 | DeepSeek V4 | 78%±9.8 | 72% | 48 |
| 22 | Kimi K3 | 73%±7.5 | 74% | 94 |
| 23 | GLM 5 | 78%±12.4 | – | 30 |
| 24 | GPT-5.3 Codex | 74%±9.8 | – | 54 |
| 25 | Kimi K2.5 | 74%±11.3 | – | 41 |
| 26 | GPT-5.6 | 76%±12.2 | 75% | 33 |
| 27 | Qwen 3 Coder Next | 72%±10.7 | – | 48 |
| 28 | Qwen 3.8 27B | 69%±9.1 | 69% | 70 |
| 29 | GLM 5.3 Flash | 71%±11.5 | 71% | 42 |
| 30 | GPT-5.2 | 68%±8.5 | – | 82 |
| 31 | Qwen 3.6 35B A3B | 70%±11.9 | 56% | 40 |
| 32 | Kimi K2.6 | 70%±12.2 | – | 38 |
| 33 | Claude Fable 5.1 | 69%±12.9 | 69% | 35 |
| 34 | Claude Sonnet 4.6 | 65%±9.7 | – | 65 |
| 35 | Gemma 4 | 62%±9.7 | – | 68 |
| 36 | Claude Sonnet 4.5 | 57%±8.3 | – | 97 |
| 37 | Qwen 3.5 | 61%±12.5 | – | 41 |
| 38 | Claude Sonnet 5 | 61%±13 | 41% | 38 |
| 39 | GPT-5 | 57%±10.3 | – | 62 |
| 40 | Gemini 3 | 56%±9.8 | – | 70 |
| 41 | GLM 4.7 | 56%±12.5 | – | 43 |
| 42 | Gemini 3.1 Pro | 54%±13.7 | – | 36 |
| 43 | Gemini 3 Pro | 52%±12.7 | – | 42 |
| 44 | GPT-6 Astra | 51%±11.9 | 51% | 48 |
| 45 | Qwen 3.8 | 50%±11.3 | 50% | 53 |
1485%1587%1685%1780%1878%1976%2076%2178%2273%2378%2474%2574%2676%2772%2869%2971%3068%3170%3270%3369%3465%3562%3657%3761%3861%3957%4056%4156%4254%4352%4451%4550%
GLM 5.3
Qwen 3 Coder
Gemini 3 Flash
DeepSeek V4.1 Flash
GLM 5.1
Qwen 3.6
Qwen 3.6 27B
DeepSeek V4
Kimi K3
GLM 5
GPT-5.3 Codex
Kimi K2.5
GPT-5.6
Qwen 3 Coder Next
Qwen 3.8 27B
GLM 5.3 Flash
GPT-5.2
Qwen 3.6 35B A3B
Kimi K2.6
Claude Fable 5.1
Claude Sonnet 4.6
Gemma 4
Claude Sonnet 4.5
Qwen 3.5
Claude Sonnet 5
GPT-5
Gemini 3
GLM 4.7
Gemini 3.1 Pro
Gemini 3 Pro
GPT-6 Astra
Qwen 3.8
Small samples: read the opinion count before the percentage.
The Shift Report
Get the monthly report
Which AI coding tools gained or lost developer approval on Hacker News, and why people switched. One email a month.