Skip to content

Benchmark

GPQA Diamond

Science benchmarks measure how well models answer science questions. Scores are percentages. Higher is better.

Models and systems measured
282
Scores recorded
1623
Sources
4
IQ weight
1.72

Score distribution

79.3%
Mean
84.5%
Median
15.7%
Std dev
56.8%
10th percentile
92.2%
90th percentile
0.168%
Min
0.168%96.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.72 for chat models, measured over 272 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 system
451Claude Fable 5.1OpenRouter harness, anthropic/claude-fable-5.1-20260831, Anthropic, public 90-day aggregate· OpenRouter benchmark exports88.9%
452GPT-6 LunaOpenRouter harness, openai/gpt-6-luna-pro-20260922, Azure, public 90-day aggregate· OpenRouter benchmark exports88.9%
453Kimi K2.7 CodeOpenRouter harness, moonshotai/kimi-k2.7-code-20260612, Moonshot AI, public 90-day aggregate· OpenRouter benchmark exports88.9%
454GPT-5.6 TerraOpenRouter harness, openai/gpt-5.6-terra-20260709, OpenAI, public 90-day aggregate· OpenRouter benchmark exports88.9%
455DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Nebius, public 90-day aggregate· OpenRouter benchmark exports88.9%
456Nova MicroOpenRouter harness, amazon/nova-micro-v1, Anthropic, public 90-day aggregate· OpenRouter benchmark exports88.9%
457Gemini 3.5 Flashlow· Epoch AI Benchmarking Hub data88.9%
458GPT 5.4medium· Epoch AI Benchmarking Hub data88.9%
459Kimi K2.5OpenRouter harness, moonshotai/kimi-k2.5-0127, Venice, public 90-day aggregate· OpenRouter benchmark exports88.8%
460Grok 4.3high· Epoch AI Benchmarking Hub data88.8%
461DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Crusoe, public 90-day aggregate· OpenRouter benchmark exports88.8%
462Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, Parasail, public 90-day aggregate· OpenRouter benchmark exports88.8%
463DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports88.8%
464Claude Opus 4.664k· Epoch AI Benchmarking Hub data88.8%
465DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Google, public 90-day aggregate· OpenRouter benchmark exports88.8%
466Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, CoreWeave, public 90-day aggregate· OpenRouter benchmark exports88.8%
467Kimi K2.7 CodeOpenRouter harness, moonshotai/kimi-k2.7-code-20260612, Nebius, public 90-day aggregate· OpenRouter benchmark exports88.8%
468DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, NextBit, public 90-day aggregate· OpenRouter benchmark exports88.8%
469Claude Fable 5.1OpenRouter harness, anthropic/claude-fable-5.1-20260831, Azure, public 90-day aggregate· OpenRouter benchmark exports88.7%
470DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Ionstream, public 90-day aggregate· OpenRouter benchmark exports88.7%
471DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Relace, public 90-day aggregate· OpenRouter benchmark exports88.7%
472Claude Fable 5OpenRouter harness, anthropic/claude-5-fable-20260609, auto-routing, public 90-day aggregate· OpenRouter benchmark exports88.7%
473Kimi K2.7 CodeOpenRouter harness, moonshotai/kimi-k2.7-code-20260612, Inceptron, public 90-day aggregate· OpenRouter benchmark exports88.7%
474Nova MicroOpenRouter harness, amazon/nova-micro-v1, auto-routing, public 90-day aggregate· OpenRouter benchmark exports88.6%
475Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, BaseTen, public 90-day aggregate· OpenRouter benchmark exports88.6%
476DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, StreamLake, public 90-day aggregate· OpenRouter benchmark exports88.6%
477Muse Spark Open 816B A42Bxhigh· Epoch AI Benchmarking Hub data88.6%
478DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Sail Research, public 90-day aggregate· OpenRouter benchmark exports88.6%
479GPT-5.6 LunaOpenRouter harness, openai/gpt-5.6-luna-20260709, Azure, public 90-day aggregate· OpenRouter benchmark exports88.6%
480DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, River, public 90-day aggregate· OpenRouter benchmark exports88.6%
481GLM 5.3OpenRouter harness, z-ai/glm-5.3-20260816, Doubleword, public 90-day aggregate· OpenRouter benchmark exports88.6%
482Qwen3.8 FlashOpenRouter harness, qwen/qwen3.8-flash-20260826, auto-routing, public 90-day aggregate· OpenRouter benchmark exports88.6%
483DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Inceptron, public 90-day aggregate· OpenRouter benchmark exports88.5%
484DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports88.5%
485Inkling Smallxhigh· Epoch AI Benchmarking Hub data88.5%
486DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, DekaLLM, public 90-day aggregate· OpenRouter benchmark exports88.5%
487GLM 5.2OpenRouter harness, z-ai/glm-5.2-20260616, Venice, public 90-day aggregate· OpenRouter benchmark exports88.5%
488DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Crusoe, public 90-day aggregate· OpenRouter benchmark exports88.5%
489Claude Fable 5OpenRouter harness, anthropic/claude-5-fable-20260609, Anthropic, public 90-day aggregate· OpenRouter benchmark exports88.5%
490Gemini 3 FlashOpenRouter harness, google/gemini-3-flash-preview-20251217, auto-routing, public 90-day aggregate· OpenRouter benchmark exports88.5%
491MiniMax M3.1 FlashOpenRouter harness, minimax/minimax-m3.1-flash-20261019, Inferact vLLM, public 90-day aggregate· OpenRouter benchmark exports88.5%
492DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Nebius, public 90-day aggregate· OpenRouter benchmark exports88.5%
493Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, CoreWeave, public 90-day aggregate· OpenRouter benchmark exports88.4%
494DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Morph, public 90-day aggregate· OpenRouter benchmark exports88.4%
495DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Nebius, public 90-day aggregate· OpenRouter benchmark exports88.4%
496Claude Fable 5OpenRouter harness, anthropic/claude-5-fable-20260609, Claude Platform on AWS, public 90-day aggregate· OpenRouter benchmark exports88.4%
497DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Novita, public 90-day aggregate· OpenRouter benchmark exports88.4%
498Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, PrimaLabs, public 90-day aggregate· OpenRouter benchmark exports88.4%
499Kimi K2.6Kimi K2.6 official launch blog · headline comparison table· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.88.4%
500Qwen3.5 397B A17B· OpenEvals Leaderboard data· Reported by the model's developer. The source relaying it did not verify it.
Values from 2 sources
88.4%

Showing 451–500 of 1,621 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 system
MiMo V2 Flash Baseofficial Base table · 5-shot· Reported by Xiaomi MiMo· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.55.1%
Qwen3.8-2.4T-A95Blifecycle-page core-benchmark claim· Reported by Alibaba Cloud Model Studio· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.92.6%

Tokens the model generated per task, as measured by the source. Comparable within this benchmark only.

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · 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. · OpenRouter public benchmark leaderboard and search-lane exports, CC BY 4.0. Fractional source scores converted to percentages; immutable dated snapshots. (licence) · OpenEvals/leaderboard-data, MIT (licence)