Skip to content

Benchmark

GDPval-AA v2

Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are Elo ratings. Higher ratings are better.

Models and systems measured
36
Scores recorded
82
Sources
1
IQ weight
2.72

Score distribution

1,587.2
Mean
1,631
Median
253.9
Std dev
1,314
10th percentile
1,844
90th percentile
735
Min
7351,932
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.72 for chat models, measured over 34 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
51DeepSeek V4 Pro 0813Artificial Analysis / Z.ai转载· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,590
52Claude Opus 4.8Artificial Analysis / Z.ai转载· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,588
53GPT-5.6 LunaGemini 3.6 Flash Model Card · July 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,584
54GPT-5.6 Terramax· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,577
55GPT-5.6 TerraArtificial Analysis / Z.ai转载· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,571
56Gemini 3.8 FlashGoogle Model Card· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,545
57Grok 4.5Gemini 3.6 Flash Model Card · July 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,535
58Gemini 3.7 FlashArtificial Analysis / Z.ai转载· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,527
59Grok 4.5high· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,526
60Gemini 3.7 Flashsetting not named by the source· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,525
61GPT 5.5xhigh · Artificial Analysis leaderboard· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,494
62GPT 5.5OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,493.7
63GPT-6 Lunaadaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · Artificial Analysis · Haiku 5.5 card snapshot · max· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,437
64Gemini 3.8 Flashadaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · Artificial Analysis · Haiku 5.5 card snapshot · highest thinking· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,435
65Gemini 3.6 Flashhigh· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,423
66Gemini 3.6 FlashGemini 3.6 Flash Model Card · July 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,421
67Claude Sonnet 4.6Artificial Analysis · as of 2026-06-17· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,395
68Muse Spark 1.1Meta Model API · xhigh · Artificial Analysis leaderboard· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,381
69Muse Spark 1.1Meta 官方发布图· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,371
70Gemini 3.5 FlashGemini 3.6 Flash Model Card · July 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,349
71Gemini 3.5 FlashOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,348.8
72Gemini 3.1 Pro PreviewFable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,314
73Gemini 3.1 Pro PreviewOpus 4.8 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,314
74Claude Haiku 5.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · Artificial Analysis · medium· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,277
75Gemini 3.1 Pro PreviewGemini 3.6 Flash Model Card · July 2026· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.965
76Gemini 3.1 Pro Previewhigh · Artificial Analysis leaderboard· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.963
77Gemini 3.1 Pro PreviewOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.962.3
78Muse Glimmer 30Bhigh reasoning· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.953
79Nemotron 3.5 Lightning 30B A3BGDPval-AA-V2 · official release table· Reported by NVIDIA· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.832
80Claude Haiku 4.5adaptive thinking · max effort · default sampling · 5-trial mean · context ≤1M unless noted · Artificial Analysis · published setting· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.735

Showing 51–80 of 80 results.

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
GPT-6 Sol (Sonnet 5.5 table label)source table label · max effort· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,487
Muse Spark (version unspecified)Meta Model API · highest effort · version unspecified · Artificial Analysis leaderboard· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.1,145

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.