GDPval-AA
We show this table for reference; we do not rank on it.
Data verified
An agentic real-world work-task evaluation reported as an Elo score in DeepSeek-V4 thinking-mode evaluations.
Benchmark score on GDPval-AA — November 13, 2026
We compile the GDPval-AA rows from provider self-reports and secondary reports. Claude Opus 5 leads the table at 1862, followed by Kimi K3.1 Pro (1861) and Muse Wonder Full (1857). We do not use these results to rank models overall.
1Closed
Claude Opus 5
Anthropic
Reported by Anthropic1862
Overall 76.05Context 1M
2Open weights
Kimi K3.1 Pro
Moonshot AI
Carried from an upstream page1861
Overall 70.36Context 1.05M
3Closed
Muse Wonder Full
Meta
Carried from an upstream page1857
Overall 72.73Context 1M
127 modelsAgenticCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (127 models)
None of the 127 results on this page has a documented run by the benchmark or an independent evaluator: 34 reported by the labs that built the models, 1 reported in another publisher’s document and 92 carried from an upstream page, where who ran them is not recorded.
127 of 127 rows
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Claude Opus 5AnthropicReported by Anthropic | 1862 | Reported by Anthropic | Not reported | Closed |
| 2 | Kimi K3.1 ProMoonshot AICarried from an upstream page | 1861 | Carried from an upstream page | Not reported | Open |
| 3 | Muse Wonder FullMetaCarried from an upstream page | 1857 | Carried from an upstream page | Not reported | Closed |
| 4 | Claude Opus 5.5AnthropicReported by Anthropic | 1846 | Reported by Anthropic | Not reported | Closed |
| 5 | Claude Sonnet 5.5AnthropicReported by Anthropic | 1844 | Reported by Anthropic | Not reported | Closed |
| 6 | Claude Fable 5.2AnthropicReported by Anthropic | 1831 | Reported by Anthropic | Not reported | Closed |
| 7 | GLM-5.3-FlashZ.AIReported by Z.AI | 1773 | Reported by Z.AI | Not reported | Open |
| 8 | Muse WonderMetaCarried from an upstream page | 1770 | Carried from an upstream page | Not reported | Closed |
| 9 | GLM-5.3Z.AIReported by Z.AI | 1769 | Reported by Z.AI | Not reported | Open |
| 10 | Claude Fable 5.1AnthropicCarried from an upstream page | 1763 | Carried from an upstream page | Not reported | Closed |
| 11 | Muse Spark 1.3MetaReported by Meta | 1754 | Reported by Meta | Not reported | Closed |
| 12 | Claude Fable 5AnthropicReported by Anthropic | 1747 | Reported by Anthropic | Not reported | Closed |
| 13 | GPT-5.6 SolOpenAIReported by OpenAI | 1735 | Reported by OpenAI | Not reported | Closed |
| 14 | Gemini 4 CarbonGoogleCarried from an upstream page | 1733 | Carried from an upstream page | Not reported | Closed |
| 15 | Grok 4.7xAICarried from an upstream page | 1717 | Carried from an upstream page | Not reported | Closed |
| 16 | Hy4 previewTencentReported by Tencent | 1678 | Reported by Tencent | Not reported | Open |
| 17 | Ling 3.1 FlashInclusionAIReported by InclusionAI | 1673 | Reported by InclusionAI | Not reported | Open |
| 18 | MiMo-V2.6-ProXiaomiReported by Xiaomi | 1673 | Reported by Xiaomi | Not reported | Open |
| 19 | Qwen3.8-Flash-NextAlibabaCarried from an upstream page | 1648 | Carried from an upstream page | Not reported | Open |
| 20 | Grok 4.6xAICarried from an upstream page | 1643 | Carried from an upstream page | Not reported | Closed |
| 21 | Muse Spark 1.2MetaReported by Meta | 1631 | Reported by Meta | Not reported | Closed |
| 22 | Inkling 2 SmallThinking Machines LabCarried from an upstream page | 1630 | Carried from an upstream page | Not reported | Closed |
| 23 | Qwen3.8 Max PreviewAlibabaCarried from an upstream page | 1630 | Carried from an upstream page | Not reported | Closed |
| 24 | Claude Haiku 5.5AnthropicReported by Anthropic | 1620 | Reported by Anthropic | Not reported | Closed |
| 25 | GPT-6.1 SolOpenAICarried from an upstream page | 1608 | Carried from an upstream page | Not reported | Closed |
| 26 | Claude Sonnet 5AnthropicReported by Anthropic | 1603 | Reported by Anthropic | Not reported | Closed |
| 27 | Claude Opus 4.8AnthropicReported by Anthropic | 1593 | Reported by Anthropic | Not reported | Closed |
| 28 | GPT-6 AstraOpenAICarried from an upstream page | 1586 | Carried from an upstream page | Not reported | Closed |
| 29 | Kimi K3Moonshot AICarried from an upstream page | 1584 | Carried from an upstream page | Not reported | Open |
| 30 | GPT-5.6 TerraOpenAIReported by OpenAI | 1583 | Reported by OpenAI | Not reported | Closed |
| 31 | Atria Dawn PreviewShanghai Artificial Intelligence LaboratoryReported by the lab | 1583 | Reported by the lab | Not reported | Open |
| 32 | GPT-5.6 LunaOpenAIReported by OpenAI | 1582 | Reported by OpenAI | Not reported | Closed |
| 33 | Step 5 PreviewStepFunReported by StepFun | 1566 | Reported by StepFun | Not reported | Open |
| 34 | Gemini 3.8 FlashGoogleReported by Google | 1545 | Reported by Google | Not reported | Closed |
| 35 | Gemini 3.7 FlashGoogleReported by Google | 1525 | Reported by Google | Not reported | Closed |
| 36 | GPT-6.1 TerraOpenAICarried from an upstream page | 1476 | Carried from an upstream page | Not reported | Closed |
| 37 | GPT-6 LunaOpenAICarried from an upstream page | 1444 | Carried from an upstream page | Not reported | Closed |
| 38 | Grok 4.5xAICarried from an upstream page | 1430 | Carried from an upstream page | Not reported | Closed |
| 39 | Qwen3.8-27BAlibabaCarried from an upstream page | 1427 | Carried from an upstream page | Not reported | Open |
| 40 | Gemini 3.6 FlashGoogleReported by Google | 1423 | Reported by Google | Not reported | Closed |
| 41 | GLM-5.2Z.AICarried from an upstream page | 1418 | Carried from an upstream page | Not reported | Open |
| 42 | GPT-5.5OpenAICarried from an upstream page | 1396 | Carried from an upstream page | Not reported | Closed |
| 43 | Claude Opus 4.7 (Adaptive)AnthropicCarried from an upstream page | 1396 | Carried from an upstream page | Not reported | Closed |
| 44 | Muse Spark 1.1MetaReported by Meta | 1375 | Reported by Meta | Not reported | Closed |
| 45 | Gemini 3.5 FlashGoogleReported by Google | 1345 | Reported by Google | Not reported | Closed |
| 46 | GPT-5.4OpenAICarried from an upstream page | 1307 | Carried from an upstream page | Not reported | Closed |
| 47 | DeepSeek V4 Pro 0813DeepSeekReported by DeepSeek | 1306 | Reported by DeepSeek | Not reported | Open |
| 48 | MiniMax M3MiniMaxCarried from an upstream page | 1304 | Carried from an upstream page | Not reported | Open |
| 49 | DeepSeek V4 Pro (High)DeepSeekCarried from an upstream page | 1294 | Carried from an upstream page | Not reported | Open |
| 50 | Ling 3.0 Flash SanteInclusionAIReported by InclusionAI | 1288 | Reported by InclusionAI | Not reported | Open |
| 51 | Apodex 1.1ApodexCarried from an upstream page | 1273 | Carried from an upstream page | Not reported | Closed |
| 52 | Apodex 1.1 MiniApodexCarried from an upstream page | 1273 | Carried from an upstream page | Not reported | Open |
| 53 | MiMo-V2.5-ProXiaomiReported by Xiaomi | 1265 | Reported by Xiaomi | Not reported | Closed |
| 54 | Quasar 438BMultiverse ComputingCarried from an upstream page | 1239 | Carried from an upstream page | Not reported | Closed |
| 55 | Nemotron 3.5 Super VL 120B A12BNVIDIAReported by NVIDIA | 1211 | Reported by NVIDIA | Not reported | Open |
| 56 | Nemotron 3.5 Super VL 120B A12B NVFP4NVIDIAReported by NVIDIA | 1206 | Reported by NVIDIA | Not reported | Open |
| 57 | Inkling-SmallThinking Machines LabCarried from an upstream page | 1191 | Carried from an upstream page | Not reported | Open |
| 58 | Qwen3.7 MaxAlibabaCarried from an upstream page | 1190 | Carried from an upstream page | Not reported | Closed |
| 59 | DeepSeek V4 Flash 0731DeepSeekReported by DeepSeek | 1189 | Reported by DeepSeek | Not reported | Open |
| 60 | GLM-5.1Z.AICarried from an upstream page | 1181 | Carried from an upstream page | Not reported | Open |
| 61 | Beam 1.0 501B-A23BReflection AICarried from an upstream page | 1177 | Carried from an upstream page | Not reported | Open |
| 62 | InklingThinking Machines LabCarried from an upstream page | 1165 | Carried from an upstream page | Not reported | Open |
| 63 | Gemini 3.5 Flash-LiteGoogleReported by Google | 1139 | Reported by Google | Not reported | Closed |
| 64 | Hy3TencentCarried from an upstream page | 1136 | Carried from an upstream page | Not reported | Open |
| 65 | Hy3 PreviewTencentCarried from an upstream page | 1136 | Carried from an upstream page | Not reported | Open |
| 66 | Kimi K2.6Moonshot AICarried from an upstream page | 1115 | Carried from an upstream page | Not reported | Open |
| 67 | Kimi K2.7 CodeMoonshot AICarried from an upstream page | 1114 | Carried from an upstream page | Not reported | Open |
| 68 | Ling 3.0 FlashInclusionAIReported by InclusionAI | 1107 | Reported by InclusionAI | Not reported | Open |
| 69 | GLM-4.7Z.AICarried from an upstream page | 1096 | Carried from an upstream page | Not reported | Open |
| 70 | GPT-5.4 miniOpenAICarried from an upstream page | 1095 | Carried from an upstream page | Not reported | Closed |
| 71 | Nemotron 3 UltraNVIDIACarried from an upstream page | 1091 | Carried from an upstream page | Not reported | Open |
| 72 | MiniMax M2.7MiniMaxCarried from an upstream page | 1087 | Carried from an upstream page | Not reported | Open |
| 73 | Muse SparkMetaCarried from an upstream page | 1076 | Carried from an upstream page | Not reported | Closed |
| 74 | Qwen3.6-27BAlibabaCarried from an upstream page | 1069 | Carried from an upstream page | Not reported | Open |
| 75 | Qwen3.6 PlusAlibabaCarried from an upstream page | 1066 | Carried from an upstream page | Not reported | Closed |
| 76 | GPT-5.4 nanoOpenAICarried from an upstream page | 1035 | Carried from an upstream page | Not reported | Closed |
| 77 | Grok 4.3xAICarried from an upstream page | 1018 | Carried from an upstream page | Not reported | Closed |
| 78 | GPT-5 (high)OpenAICarried from an upstream page | 1015 | Carried from an upstream page | Not reported | Closed |
| 79 | Qwen3.6-35B-A3BAlibabaCarried from an upstream page | 992 | Carried from an upstream page | Not reported | Open |
| 80 | Step 3.7 FlashStepFunCarried from an upstream page | 954 | Carried from an upstream page | Not reported | Open |
| 81 | Kimi K2.5 (Reasoning)Moonshot AICarried from an upstream page | 936 | Carried from an upstream page | Not reported | Open |
| 82 | Kimi K2.5Moonshot AICarried from an upstream page | 936 | Carried from an upstream page | Not reported | Open |
| 83 | GPT-5.1OpenAICarried from an upstream page | 930 | Carried from an upstream page | Not reported | Closed |
| 84 | Qwen3.5-122B-A10BAlibabaCarried from an upstream page | 925 | Carried from an upstream page | Not reported | Open |
| 85 | Gemini 3.1 ProGoogleCarried from an upstream page | 904 | Carried from an upstream page | Not reported | Closed |
| 86 | Muse Glimmer 30BMetaCarried from an upstream page | 893 | Carried from an upstream page | Not reported | Open |
| 87 | Qwen3.7 PlusAlibabaCarried from an upstream page | 886 | Carried from an upstream page | Not reported | Closed |
| 88 | GPT-5 miniOpenAICarried from an upstream page | 880 | Carried from an upstream page | Not reported | Closed |
| 89 | Mistral Medium 3.5 128BMistralCarried from an upstream page | 875 | Carried from an upstream page | Not reported | Open |
| 90 | Nemotron 3.5 Lightning 30B A3B NVFP4NVIDIAReported by NVIDIA | 865 | Reported by NVIDIA | Not reported | Open |
| 91 | MiMo-V2-FlashXiaomiCarried from an upstream page | 788 | Carried from an upstream page | Not reported | Open |
| 92 | Gemma 4 31BGoogleCarried from an upstream page | 755 | Carried from an upstream page | Not reported | Open |
| 93 | GPT-OSS 120BOpenAICarried from an upstream page | 745 | Carried from an upstream page | Not reported | Open |
| 94 | Gemma 4 26B A4BGoogleCarried from an upstream page | 713 | Carried from an upstream page | Not reported | Open |
| 95 | Command A+CohereCarried from an upstream page | 658 | Carried from an upstream page | Not reported | Open |
| 96 | Granite 4.2 8BIBMCarried from an upstream page | 646 | Carried from an upstream page | Not reported | Open |
| 97 | Mercury 2InceptionCarried from an upstream page | 646 | Carried from an upstream page | Not reported | Closed |
| 98 | Gemini 2.5 ProGoogleCarried from an upstream page | 616 | Carried from an upstream page | Not reported | Closed |
| 99 | Gemma 4 12BGoogleCarried from an upstream page | 591 | Carried from an upstream page | Not reported | Open |
| 100 | Mistral Large 3MistralCarried from an upstream page | 588 | Carried from an upstream page | Not reported | Open |
| 101 | Ling 2.6 FlashInclusionAIReported by a third party | 550 | Reported by a third party | Not reported | Open |
| 102 | K-ExaoneLG AI ResearchCarried from an upstream page | 540 | Carried from an upstream page | Not reported | Closed |
| 103 | Mistral Small 4MistralCarried from an upstream page | 537 | Carried from an upstream page | Not reported | Open |
| 104 | Mistral Small 4 (Reasoning)MistralCarried from an upstream page | 537 | Carried from an upstream page | Not reported | Open |
| 105 | GPT-OSS 20BOpenAICarried from an upstream page | 514 | Carried from an upstream page | Not reported | Open |
| 106 | Trinity-Large-ThinkingArcee AICarried from an upstream page | 512 | Carried from an upstream page | Not reported | Open |
| 107 | Trinity-Large-PreviewArcee AICarried from an upstream page | 512 | Carried from an upstream page | Not reported | Open |
| 108 | Celeris-1CelerisCarried from an upstream page | 483 | Carried from an upstream page | Not reported | Closed |
| 109 | GPT-4.1 miniOpenAICarried from an upstream page | 453 | Carried from an upstream page | Not reported | Closed |
| 110 | Nemotron 3 Nano 30BNVIDIACarried from an upstream page | 438 | Carried from an upstream page | Not reported | Open |
| 111 | Ministral 3 14B (Reasoning)MistralCarried from an upstream page | 432 | Carried from an upstream page | Not reported | Open |
| 112 | Ministral 3 14BMistralCarried from an upstream page | 432 | Carried from an upstream page | Not reported | Open |
| 113 | Nemotron 3 Nano Omni 30B A3BNVIDIACarried from an upstream page | 416 | Carried from an upstream page | Not reported | Open |
| 114 | Ministral 3 8B (Reasoning)MistralCarried from an upstream page | 405 | Carried from an upstream page | Not reported | Open |
| 115 | Ministral 3 8BMistralCarried from an upstream page | 405 | Carried from an upstream page | Not reported | Open |
| 116 | Ministral 3 3B (Reasoning)MistralCarried from an upstream page | 234 | Carried from an upstream page | Not reported | Open |
| 117 | Ministral 3 3BMistralCarried from an upstream page | 234 | Carried from an upstream page | Not reported | Open |
| 118 | LFM2.5-2.6BLiquidAICarried from an upstream page | 202 | Carried from an upstream page | Not reported | Open |
| 119 | GPT-4o miniOpenAICarried from an upstream page | 188 | Carried from an upstream page | Not reported | Closed |
| 120 | DeepSeek V3DeepSeekCarried from an upstream page | 183 | Carried from an upstream page | Not reported | Open |
| 121 | Gemma 4 E4BGoogleCarried from an upstream page | 177 | Carried from an upstream page | Not reported | Open |
| 122 | Llama 4 ScoutMetaCarried from an upstream page | 60 | Carried from an upstream page | Not reported | Open |
| 123 | Ultravox v0.6 Llama 3.3 70BFixie AICarried from an upstream page | 50 | Carried from an upstream page | Not reported | Open |
| 124 | Gemma 4 E2BGoogleCarried from an upstream page | 36 | Carried from an upstream page | Not reported | Open |
| 125 | GPT-4.1 nanoOpenAICarried from an upstream page | 13 | Carried from an upstream page | Not reported | Closed |
| 126 | Llama 4 MaverickMetaCarried from an upstream page | -45 | Carried from an upstream page | Not reported | Open |
| 127 | Gemma 3 27BGoogleCarried from an upstream page | -171 | Carried from an upstream page | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
Among the reported GDPval-AA rows, Claude Opus 5 is first at 1862. The third row is 5 score units behind. The broader top-10 range is 99 score units, so the table still separates the published systems.
127 models have been evaluated on GDPval-AA. The benchmark falls in the Agentic category. GDPval-AA is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About GDPval-AA
Year
2026
Tasks
Agentic real-world work tasks
Format
Elo
Difficulty
Professional agentic workflows
BenchLM stores GDPval-AA as a display-only provider-table row for DeepSeek-V4 because the source reports an Elo score rather than a 0-100 percentage.
Freshness and provenance
Version
GDPval-AA 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
CurrentDisplay only
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 GDPval-AA measure?
An agentic real-world work-task evaluation reported as an Elo score in DeepSeek-V4 thinking-mode evaluations.
Which model scores highest on GDPval-AA?
Claude Opus 5 by Anthropic currently leads with a score of 1862 on GDPval-AA.
How many models are evaluated on GDPval-AA?
127 AI models have published results on GDPval-AA in the BenchLM catalog.
Last updated: November 13, 2026 · BenchLM version GDPval-AA 2026
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 7,000+ subscribers.
One email each week. Unsubscribe anytime.