Artificial Analysis Coding Index (AA Coding Index)
We show this table for reference; we do not rank on it.
A display-only Artificial Analysis coding index.
Benchmark score on AA Coding Index — October 29, 2026
We compile the AA Coding Index rows from provider self-reports and secondary reports. Claude Fable 5.1 leads the table at 81.6%, followed by Claude Opus 5 (78.0%) and GPT-5.6 Sol (77.4%). We do not use these results to rank models overall.
Claude Fable 5.1
Anthropic
Claude Opus 5
Anthropic
GPT-5.6 Sol
OpenAI
109 modelsCodingCurrentDisplay onlyUpdated October 29, 2026
Benchmark score table (109 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Claude Fable 5.1Anthropic | 81.6% | Not reported | Closed |
| 2 | Claude Opus 5Anthropic | 78.0% | Not reported | Closed |
| 3 | GPT-5.6 SolOpenAI | 77.4% | Not reported | Closed |
| 4 | GPT-6 AstraOpenAI | 76.9% | Not reported | Closed |
| 5 | Grok 4.6xAI | 76.8% | Not reported | Closed |
| 6 | GPT-5.6 TerraOpenAI | 76.7% | Not reported | Closed |
| 7 | Claude Fable 5Anthropic | 76.5% | Not reported | Closed |
| 8 | Gemini 3.8 FlashGoogle | 76.3% | Not reported | Closed |
| 9 | Kimi K3Moonshot AI | 76.2% | Not reported | Open |
| 10 | Gemini 3.7 FlashGoogle | 76.1% | Not reported | Closed |
| 11 | Muse Spark 1.3Meta | 75.8% | Not reported | Closed |
| 12 | GPT-5.5OpenAI | 74.9% | Not reported | Closed |
| 13 | GLM-5.3Z.AI | 74.8% | Not reported | Open |
| 14 | Claude Opus 4.8Anthropic | 74.3% | Not reported | Closed |
| 15 | Claude Opus 4.7 (Adaptive)Anthropic | 73.6% | Not reported | Closed |
| 16 | Qwen3.8-Flash-NextAlibaba | 73.0% | Not reported | Open |
| 17 | Grok 4.5xAI | 72.5% | Not reported | Closed |
| 18 | Muse Spark 1.2Meta | 72.2% | Not reported | Closed |
| 19 | Qwen3.8 Max PreviewAlibaba | 71.8% | Not reported | Closed |
| 20 | Claude Sonnet 5Anthropic | 71.5% | Not reported | Closed |
| 21 | GPT-5.6 LunaOpenAI | 71.5% | Not reported | Closed |
| 22 | Muse Spark 1.1Meta | 71.3% | Not reported | Closed |
| 23 | GPT-5.4OpenAI | 71.0% | Not reported | Closed |
| 24 | Gemini 3.5 FlashGoogle | 70.1% | Not reported | Closed |
| 25 | Gemini 3.6 FlashGoogle | 69.2% | Not reported | Closed |
| 26 | DeepSeek V4 Flash 0731DeepSeek | 69.1% | Not reported | Open |
| 27 | Gemini 3.1 ProGoogle | 68.8% | Not reported | Closed |
| 28 | DeepSeek V4 Pro 0813DeepSeek | 68.8% | Not reported | Open |
| 29 | GLM-5.2Z.AI | 68.8% | Not reported | Open |
| 30 | Qwen3.8-27BAlibaba | 68.1% | Not reported | Open |
| 31 | Qwen3.7 MaxAlibaba | 66.0% | Not reported | Closed |
| 32 | Kimi K2.6Moonshot AI | 61.8% | Not reported | Open |
| 33 | Quasar 438BMultiverse Computing | 61.2% | Not reported | Closed |
| 34 | Kimi K2.7 CodeMoonshot AI | 60.8% | Not reported | Open |
| 35 | Apodex 1.1Apodex | 60.8% | Not reported | Closed |
| 36 | Apodex 1.1 MiniApodex | 60.8% | Not reported | Open |
| 37 | MiMo-V2.5-ProXiaomi | 60.2% | Not reported | Closed |
| 38 | Hy3Tencent | 58.8% | Not reported | Open |
| 39 | Hy3 PreviewTencent | 58.8% | Not reported | Open |
| 40 | DeepSeek V4 Pro (High)DeepSeek | 58.7% | Not reported | Open |
| 41 | Muse SparkMeta | 58.6% | Not reported | Closed |
| 42 | MiniMax M3MiniMax | 58.6% | Not reported | Open |
| 43 | GPT-5.4 miniOpenAI | 56.1% | Not reported | Closed |
| 44 | GPT-5.4 nanoOpenAI | 56.1% | Not reported | Closed |
| 45 | Qwen3.7 PlusAlibaba | 55.9% | Not reported | Closed |
| 46 | GLM-5.1Z.AI | 55.8% | Not reported | Open |
| 47 | Qwen3.6 PlusAlibaba | 54.5% | Not reported | Closed |
| 48 | Qwen3.6-27BAlibaba | 53.7% | Not reported | Open |
| 49 | Inkling-SmallThinking Machines Lab | 53.0% | Not reported | Open |
| 50 | MiniMax M2.7MiniMax | 52.6% | Not reported | Open |
| 51 | InklingThinking Machines Lab | 52.1% | Not reported | Open |
| 52 | Ling 3.0 FlashInclusionAI | 50.6% | Not reported | Open |
| 53 | MiMo-V2-FlashXiaomi | 49.8% | Not reported | Open |
| 54 | GPT-5.1OpenAI | 49.4% | Not reported | Closed |
| 55 | Gemini 3.5 Flash-LiteGoogle | 49.3% | Not reported | Closed |
| 56 | Nemotron 3 UltraNVIDIA | 49.3% | Not reported | Open |
| 57 | Muse Glimmer 30BMeta | 49.0% | Not reported | Open |
| 58 | Mistral Medium 3.5 128BMistral | 46.9% | Not reported | Open |
| 59 | Kimi K2.5 (Reasoning)Moonshot AI | 46.8% | Not reported | Closed |
| 60 | Kimi K2.5Moonshot AI | 46.8% | Not reported | Open |
| 61 | Qwen3.5-122B-A10BAlibaba | 45.7% | Not reported | Open |
| 62 | GLM-4.7Z.AI | 45.3% | Not reported | Open |
| 63 | Gemma 4 31BGoogle | 43.4% | Not reported | Open |
| 64 | Grok 4.3xAI | 42.3% | Not reported | Closed |
| 65 | Qwen3.6-35B-A3BAlibaba | 41.9% | Not reported | Open |
| 66 | o1OpenAI | 39.7% | Not reported | Closed |
| 67 | Step 3.7 FlashStepFun | 39.6% | Not reported | Open |
| 68 | Gemma 4 26B A4BGoogle | 39.3% | Not reported | Open |
| 69 | GPT-5 (high)OpenAI | 37.8% | Not reported | Closed |
| 70 | o1-previewOpenAI | 34.0% | Not reported | Closed |
| 71 | Gemini 2.5 ProGoogle | 33.3% | Not reported | Closed |
| 72 | K-ExaoneLG AI Research | 32.1% | Not reported | Closed |
| 73 | Mercury 2Inception | 31.1% | Not reported | Closed |
| 74 | Gemma 4 12BGoogle | 31.0% | Not reported | Open |
| 75 | GPT-OSS 120BOpenAI | 30.4% | Not reported | Open |
| 76 | Command A+Cohere | 27.9% | Not reported | Open |
| 77 | Nemotron 3.5 Lightning 30B A3B NVFP4NVIDIA | 26.8% | Not reported | Open |
| 78 | Mistral Small 4Mistral | 26.6% | Not reported | Open |
| 79 | Mistral Small 4 (Reasoning)Mistral | 26.6% | Not reported | Open |
| 80 | Trinity-Large-ThinkingArcee AI | 25.8% | Not reported | Open |
| 81 | Trinity-Large-PreviewArcee AI | 25.8% | Not reported | Open |
| 82 | Ling 2.6 FlashInclusionAI | 25.3% | Not reported | Open |
| 83 | Gemini 1.5 ProGoogle | 23.6% | Not reported | Closed |
| 84 | DeepSeek V3DeepSeek | 23.0% | Not reported | Open |
| 85 | Granite 4.2 8BIBM | 22.4% | Not reported | Open |
| 86 | GPT-4 TurboOpenAI | 21.5% | Not reported | Closed |
| 87 | GPT-OSS 20BOpenAI | 20.7% | Not reported | Open |
| 88 | GPT-4.1 miniOpenAI | 20.2% | Not reported | Closed |
| 89 | Mistral Large 3Mistral | 20.1% | Not reported | Closed |
| 90 | Claude 3 OpusAnthropic | 19.5% | Not reported | Closed |
| 91 | Llama 4 MaverickMeta | 16.3% | Not reported | Open |
| 92 | GPT-5 miniOpenAI | 15.6% | Not reported | Closed |
| 93 | Celeris-1Celeris | 14.4% | Not reported | Closed |
| 94 | Nemotron 3 Nano 30BNVIDIA | 14.4% | Not reported | Open |
| 95 | Ministral 3 14B (Reasoning)Mistral | 14.4% | Not reported | Open |
| 96 | Ministral 3 14BMistral | 14.4% | Not reported | Open |
| 97 | Nemotron 3 Nano Omni 30B A3BNVIDIA | 13.8% | Not reported | Open |
| 98 | Ultravox v0.6 Llama 3.3 70BFixie AI | 11.9% | Not reported | Open |
| 99 | GPT-4o miniOpenAI | 11.4% | Not reported | Closed |
| 100 | GPT-4.1 nanoOpenAI | 11.1% | Not reported | Closed |
| 101 | Gemma 3 27BGoogle | 10.1% | Not reported | Open |
| 102 | Ministral 3 8B (Reasoning)Mistral | 9.7% | Not reported | Open |
| 103 | Ministral 3 8BMistral | 9.7% | Not reported | Open |
| 104 | Gemma 4 E4BGoogle | 9.4% | Not reported | Open |
| 105 | Llama 4 ScoutMeta | 8.2% | Not reported | Open |
| 106 | LFM2.5-2.6BLiquidAI | 7.7% | Not reported | Open |
| 107 | Gemma 4 E2BGoogle | 7.2% | Not reported | Open |
| 108 | Ministral 3 3B (Reasoning)Mistral | 4.8% | Not reported | Open |
| 109 | Ministral 3 3BMistral | 4.8% | Not reported | Open |
Among the reported AA Coding Index rows, Claude Fable 5.1 is first at 81.6%. The third row is 4.2 points behind. The broader top-10 range is 5.5 points, so many of the published results sit in a relatively narrow band.
109 models have been evaluated on AA Coding Index. The benchmark falls in the Coding category. AA Coding Index is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About AA Coding Index
Year
2026
Tasks
Cross-benchmark coding index
Format
Aggregated model score
Difficulty
Display-only external reference
BenchLM mirrors this coding index for comparison, but does not use it as a weighted coding benchmark row.
Freshness and provenance
Version
AA Coding Index 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does AA Coding Index measure?
A display-only Artificial Analysis coding index.
Which model scores highest on AA Coding Index?
Claude Fable 5.1 by Anthropic currently leads with a score of 81.6% on AA Coding Index.
How many models are evaluated on AA Coding Index?
109 AI models have published results on AA Coding Index in the BenchLM catalog.
Compare top models on AA Coding Index
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.