Model results

GPQA Diamond

Choose a shared evaluation protocol to compare model scores. Versions, sources and configurations stay separate.

Atlas snapshot captured 10/19/2026, 10:20:28 AM UTC. This is an import timestamp, not the evaluation date. Coverage may be incomplete.

Reviewed recommendation and limits

Use as a compact reference for expert-written science questions, alongside evidence that matches the scientific work you intend to do.

Original GPQA Diamond subset. Preserve prompt, answer extraction, tools, sampling and aggregation. Keep main, extended and third-party corrected subsets separate.. Imported groups below may use different protocols; this review does not certify each run. The static 198-question subset has limited statistical power and exposure risk. Its expert filtering does not make it representative of scientific practice. A Google-proof collection process does not prove present-day contamination resistance.

Publisher results GPQA paper

← All evaluation groups

GPQA Diamond

Source: Epoch AI Benchmarking Hub
Version: Not reported
Measurement date: Not reported
Metric: % · Higher is better
Evaluation scope: Epoch AI run · task v1 · effort high · result date not supplied

Exact score ties share a position. Positions do not establish statistical significance. Missing metadata is shown as unreported; matching imported metadata is not independent verification of the protocol.

#1Gemini 4 Argon
google/gemini-4-argon-20261013
96.4%
View modelEvidence independent
#2Gemini 3.8 Flash
google/gemini-3.8-flash-20260902
95.4%
View modelEvidence independent
#3Gemini 3.7 Flash
google/gemini-3.7-flash-20260813
94.8%
View modelEvidence independent
#4Gemini 3.1 Pro Preview
google/gemini-3.1-pro-preview-20260219
94.4%
View modelEvidence independent
#5Gemini 3.6 Flash
google/gemini-3.6-flash-20260721
94.1%
View modelEvidence independent
#6Grok 4.6
x-ai/grok-4.6-20260810
94%
View modelEvidence independent
#7Grok 4.5
x-ai/grok-4.5-20260708
93.4%
View modelEvidence independent
#8Gemini 3.5 Flash
google/gemini-3.5-flash-20260519
92.8%
View modelEvidence independent
#9Kimi K3
moonshotai/kimi-k3-20260715
91.9%
View modelEvidence independent
#10DeepSeek V4 Pro 0423
deepseek/deepseek-v4-pro-20260423
90.9%
View modelEvidence independent
#11GPT-5.4
openai/gpt-5.4-20260305
89.9%
View modelEvidence independent
#12Gemini 3 Flash Preview
google/gemini-3-flash-preview-20251217
89.4%
View modelEvidence independent
#13Grok 4.3
x-ai/grok-4.3-20260430
88.8%
View modelEvidence independent
#14GPT-5.2
openai/gpt-5.2-20251211
88.2%
View modelEvidence independent
#15GPT-5.1
openai/gpt-5.1-20251113
87.6%
View modelEvidence independent
#16GPT-5
openai/gpt-5-2025-08-07
86.2%
View modelEvidence independent
#17GPT-5.4 Mini
openai/gpt-5.4-mini-20260317
83.6%
View modelEvidence independent
#18Claude Fable 5
anthropic/claude-fable-5
83.3%
View modelEvidence independent
#18Claude Sonnet 4.6
anthropic/claude-sonnet-4-6
83.3%
View modelEvidence independent
#18Gemini 3.5 Flash Lite
google/gemini-3.5-flash-lite-20260721
83.3%
View modelEvidence independent
#21Gemini 3.1 Flash Lite
google/gemini-3.1-flash-lite-20260507
81.8%
View modelEvidence independent
#21o3
openai/o3-2025-04-16
81.8%
View modelEvidence independent
#23o4 Mini
openai/o4-mini-2025-04-16
79.6%
View modelEvidence independent
#24GPT-5.4 Nano
openai/gpt-5.4-nano-20260317
78.5%
View modelEvidence independent
#25o3 Mini
openai/o3-mini-2025-01-31
77%
View modelEvidence independent
#26o1
openai/o1-2024-12-17
76.8%
View modelEvidence independent
#27gpt-oss-120b
openai/gpt-oss-120b
75.8%
View modelEvidence independent
#28GPT-5 Mini
openai/gpt-5-mini-2025-08-07
75%
View modelEvidence independent
#29GPT-5 Nano
openai/gpt-5-nano-2025-08-07
69.4%
View modelEvidence independent
#30gpt-oss-20b
openai/gpt-oss-20b
46%
View modelEvidence independent