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
401DeepSeek V4 FlashOpenRouter harness, deepseek/deepseek-v4-flash-20260423, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports89.5%
402Hy4 previewOpenRouter harness, tencent/hy4-preview-20260827, Tencent, public 90-day aggregate· OpenRouter benchmark exports89.5%
403DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, ModelRun, public 90-day aggregate· OpenRouter benchmark exports89.5%
404DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Cohere, public 90-day aggregate· OpenRouter benchmark exports89.5%
405DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, ModelRun, public 90-day aggregate· OpenRouter benchmark exports89.4%
406DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Morph, public 90-day aggregate· OpenRouter benchmark exports89.4%
407Claude Opus 4.7OpenRouter harness, anthropic/claude-4.7-opus-20260416, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports89.4%
408Hy4 previewOpenRouter harness, tencent/hy4-preview-20260827, Venice, public 90-day aggregate· OpenRouter benchmark exports89.4%
409Gemini 3 Flashhigh· Epoch AI Benchmarking Hub data89.4%
410DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, Phala, public 90-day aggregate· OpenRouter benchmark exports89.4%
411DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Ambient, public 90-day aggregate· OpenRouter benchmark exports89.4%
412Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports89.4%
413Claude Opus 4.7OpenRouter harness, anthropic/claude-4.7-opus-20260416, Anthropic, public 90-day aggregate· OpenRouter benchmark exports89.4%
414Hy4 previewOpenRouter harness, tencent/hy4-preview-20260827, Makora, public 90-day aggregate· OpenRouter benchmark exports89.3%
415Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, Baidu, public 90-day aggregate· OpenRouter benchmark exports89.3%
416Doubao Seed 2.1 Prohigh· Epoch AI Benchmarking Hub data89.3%
417Grok 4.20reasoning, 0309· Epoch AI Benchmarking Hub data89.3%
418GPT-5.6 TerraOpenRouter harness, openai/gpt-5.6-terra-20260709, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports89.3%
419Hy4 previewOpenRouter harness, tencent/hy4-preview-20260827, AtlasCloud, public 90-day aggregate· OpenRouter benchmark exports89.3%
420DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Fireworks, public 90-day aggregate· OpenRouter benchmark exports89.3%
421Kimi K2.7 CodeOpenRouter harness, moonshotai/kimi-k2.7-code-20260612, Fireworks, public 90-day aggregate· OpenRouter benchmark exports89.3%
422Hy4 previewOpenRouter harness, tencent/hy4-preview-20260827, Novita, public 90-day aggregate· OpenRouter benchmark exports89.3%
423MiniMax M3.1 FlashOpenRouter harness, minimax/minimax-m3.1-flash-20261019, Novita, public 90-day aggregate· OpenRouter benchmark exports89.3%
424Claude Opus 5OpenRouter harness, anthropic/claude-opus-5-20260723, Google, public 90-day aggregate· OpenRouter benchmark exports89.3%
425Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports89.3%
426DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, InferenceNet, public 90-day aggregate· OpenRouter benchmark exports89.3%
427Gemini 3 FlashOpenRouter harness, google/gemini-3-flash-preview-20251217, Google AI Studio, public 90-day aggregate· OpenRouter benchmark exports89.2%
428GPT-5.6 TerraOpenRouter harness, openai/gpt-5.6-terra-20260709, Azure, public 90-day aggregate· OpenRouter benchmark exports89.2%
429Qwen 3.8 27BQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.89.2%
430Qwen 3.8 27BQwen3.8-27B model card · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.89.2%
431DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Wafer, public 90-day aggregate· OpenRouter benchmark exports89.2%
432DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Mancer 2, public 90-day aggregate· OpenRouter benchmark exports89.2%
433DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Phala, public 90-day aggregate· OpenRouter benchmark exports89.2%
434DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, OpenInference, public 90-day aggregate· OpenRouter benchmark exports89.2%
435DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports89.2%
436DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Alibaba, public 90-day aggregate· OpenRouter benchmark exports89.2%
437Kimi K2.7 CodeOpenRouter harness, moonshotai/kimi-k2.7-code-20260612, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports89.1%
438DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, auto-routing, public 90-day aggregate· OpenRouter benchmark exports89.1%
439DeepSeek V4.1 FlashOpenRouter harness, deepseek/deepseek-v4.1-flash-20260910, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports89.1%
440Claude Opus 4.7OpenRouter harness, anthropic/claude-4.7-opus-20260416, Azure, public 90-day aggregate· OpenRouter benchmark exports89.1%
441DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Fireworks, public 90-day aggregate· OpenRouter benchmark exports89.1%
442DeepSeek V4 FlashOpenRouter harness, deepseek/deepseek-v4-flash-20260423, GMICloud, public 90-day aggregate· OpenRouter benchmark exports89%
443Kimi K2.7 CodeOpenRouter harness, moonshotai/kimi-k2.7-code-20260612, Novita, public 90-day aggregate· OpenRouter benchmark exports89%
444Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports89%
445GPT-6 LunaOpenRouter harness, openai/gpt-6-luna-pro-20260922, OpenAI, public 90-day aggregate· OpenRouter benchmark exports89%
446GLM 5.1OpenRouter harness, z-ai/glm-5.1-20260406, Venice, public 90-day aggregate· OpenRouter benchmark exports89%
447MiMo V2.6 ProOpenRouter harness, xiaomi/mimo-v2.6-pro-20260921, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports89%
448DeepSeek V4 Flash 0731OpenRouter harness, deepseek/deepseek-v4-flash-20260731, Reka, public 90-day aggregate· OpenRouter benchmark exports89%
449Kimi K2.6OpenRouter harness, moonshotai/kimi-k2.6-20260420, Venice, public 90-day aggregate· OpenRouter benchmark exports89%
450DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Ionstream, public 90-day aggregate· OpenRouter benchmark exports89%

Showing 401–450 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)