Reasoning · Benchmark profile

GPQA Diamond

The Diamond subset of a graduate-level, expert-written question-answering benchmark.

Ranking weightedCurrent / rollingActive85 models

Data verified 15 July 2026 · Methodology 1.2.0

Benchmark score on GPQA Diamond

Leaderboard · 85 models

GPT-5.6 Sol leads at 94.6%, followed by Gemini 3.1 Pro Preview (94.1%) and Claude Mythos 5 (94.1%).

Ranking · 2.5%ReasoningCurrent

#1Closed

GPT-5.6 Sol

OpenAI

gpt-5.6-sol

94.6%

Overall #4 · Context 1.05M tokens

#2Closed

Gemini 3.1 Pro Preview

Google

gemini-3.1-pro-preview

94.1%

Overall #42 · Context 1.05M tokens

#3Closed

Claude Mythos 5

Anthropic

claude-mythos-5

94.1%

Overall #1 · Context 1M tokens

Visual analysis

See the field, not just the table.

Three views of the same sourced leaderboard: model placement, the score curve, and descriptive provider averages. These visuals never alter ranking calculations.

ModelsColours

OpenAIGoogleAnthropicMoonshot AIxAIMiniMaxAlibaba CloudZ.AI

Leaderboard bars

20 ranked models · higher is better

View accessible chart data

Accessible data for Leaderboard bars
Model	Rank	Provider	Score
GPT-5.6 Sol	#1	OpenAI	94.6%
Gemini 3.1 Pro Preview	#2	Google	94.1%
Claude Mythos 5	#3	Anthropic	94.1%
Claude Opus 4.8	#4	Anthropic	93.6%
GPT-5.5	#5	OpenAI	93.6%
Kimi K3	#6	Moonshot AI	93.5%
Grok 4.5	#7	xAI	93.1%
MiniMax M3	#8	MiniMax	92.9%
GPT-5.6 Terra	#9	OpenAI	92.9%
GPT-5.4	#10	OpenAI	92.8%
Claude Fable 5	#11	Anthropic	92.6%
Qwen3.7-Max	#12	Alibaba Cloud	92.4%
GPT-5.6 Luna	#13	OpenAI	92.3%
Gemini 3.5 Flash	#14	Google	92.2%
GPT-5.3-Codex	#15	OpenAI	91.5%
Claude Opus 4.6	#16	Anthropic	91.3%
GLM-5.2	#17	Z.AI	91.2%
Grok 4.20	#18	xAI	91.1%
Kimi K2.6	#19	Moonshot AI	91.1%
Claude Sonnet 5	#20	Anthropic	91.1%

Score curve

Shows how quickly scores separate down the field

View accessible chart data

Accessible data for Score curve
Model	Rank	Provider	Score
GPT-5.6 Sol	#1	OpenAI	94.6%
Gemini 3.1 Pro Preview	#2	Google	94.1%
Claude Mythos 5	#3	Anthropic	94.1%
Claude Opus 4.8	#4	Anthropic	93.6%
GPT-5.5	#5	OpenAI	93.6%
Kimi K3	#6	Moonshot AI	93.5%
Grok 4.5	#7	xAI	93.1%
MiniMax M3	#8	MiniMax	92.9%
GPT-5.6 Terra	#9	OpenAI	92.9%
GPT-5.4	#10	OpenAI	92.8%
Claude Fable 5	#11	Anthropic	92.6%
Qwen3.7-Max	#12	Alibaba Cloud	92.4%
GPT-5.6 Luna	#13	OpenAI	92.3%
Gemini 3.5 Flash	#14	Google	92.2%
GPT-5.3-Codex	#15	OpenAI	91.5%
Claude Opus 4.6	#16	Anthropic	91.3%
GLM-5.2	#17	Z.AI	91.2%
Grok 4.20	#18	xAI	91.1%
Kimi K2.6	#19	Moonshot AI	91.1%
Claude Sonnet 5	#20	Anthropic	91.1%

Provider profile

Average within the visible cohort · descriptive only

View accessible chart data

Accessible data for Provider profile
Provider	Models	Average score
Google	2	93.2%
OpenAI	6	93.0%
MiniMax	1	92.9%
Anthropic	5	92.5%
Alibaba Cloud	1	92.4%
Moonshot AI	2	92.3%
xAI	2	92.1%
Z.AI	1	91.2%

Benchmark score table (85 models)

One best score per model · higher is better

GPQA Diamond leaderboard ranked by score
Rank	Model	Provider	License	Score
#1	GPT-5.6 Sol gpt-5.6-sol	OpenAI	closed	94.6%
#2	Gemini 3.1 Pro Preview gemini-3.1-pro-preview	Google	closed	94.1%
#3	Claude Mythos 5 claude-mythos-5	Anthropic	closed	94.1%
#4	Claude Opus 4.8 claude-opus-4-8	Anthropic	closed	93.6%
#5	GPT-5.5 gpt-5.5	OpenAI	closed	93.6%
#6	Kimi K3 kimi-k3	Moonshot AI	closed	93.5%
#7	Grok 4.5 grok-4.5	xAI	closed	93.1%
#8	MiniMax M3 MiniMax-M3	MiniMax	—	92.9%
#9	GPT-5.6 Terra gpt-5.6-terra	OpenAI	closed	92.9%
#10	GPT-5.4 gpt-5.4	OpenAI	closed	92.8%
#11	Claude Fable 5 claude-fable-5	Anthropic	closed	92.6%
#12	Qwen3.7-Max qwen3.7-max	Alibaba Cloud	—	92.4%
#13	GPT-5.6 Luna gpt-5.6-luna	OpenAI	closed	92.3%
#14	Gemini 3.5 Flash gemini-3.5-flash	Google	closed	92.2%
#15	GPT-5.3-Codex gpt-5.3-codex	OpenAI	closed	91.5%
#16	Claude Opus 4.6 claude-opus-4-6	Anthropic	closed	91.3%
#17	GLM-5.2 glm-5-2	Z.AI	open	91.2%
#18	Grok 4.20 grok-4.20	xAI	closed	91.1%
#19	Kimi K2.6 kimi-k2.6	Moonshot AI	open	91.1%
#20	Claude Sonnet 5 claude-sonnet-5	Anthropic	closed	91.1%
#21	Hy3 hy3	Tencent	open	90.4%
#22	Qwen3.6 Plus qwen3-6-plus	Alibaba Cloud	closed	90.4%
#23	GPT-5.2 gpt-5.2	OpenAI	closed	90.3%
#24	Qwen3.7-Plus qwen3.7-plus	Alibaba Cloud	—	90.3%
#25	Grok 4.3 grok-4.3	xAI	closed	90.1%
#26	Claude Sonnet 4.6 claude-sonnet-4-6	Anthropic	closed	89.9%
#27	GPT-5.2 Codex gpt-5.2-codex	OpenAI	closed	89.9%
#28	Muse Spark 1.1 muse-spark-1-1	Meta	closed	89.8%
#29	Kimi K2.7 Code kimi-k2.7-code	Moonshot AI	open	89.6%
#30	DeepSeek V4 Flash deepseek-v4-flash	DeepSeek	open	89.4%
#31	Qwen3.5 397B qwen3.5-397b-a17b	Alibaba Cloud	open	89.3%
#32	DeepSeek V4 Pro deepseek-v4-pro	DeepSeek	open	88.8%
#33	Qwen3.6 Max qwen3.6-max	Alibaba Cloud	closed	88.8%
#34	Grok 4 grok-4	xAI	closed	87.7%
#35	Kimi K2.5 kimi-k2.5	Moonshot AI	—	87.6%
#36	GPT-5.4 mini gpt-5.4-mini	OpenAI	closed	87.5%
#37	Claude Opus 4.5 claude-opus-4-5	Anthropic	closed	87%
#38	Nemotron 3 Ultra nemotron-3-ultra-550b	NVIDIA	open	86.7%
#39	Qwen3 Max qwen3-max	Alibaba Cloud	closed	86.1%
#40	GLM-5 glm-5	Z.AI	open	86%
#41	GLM-4.7 glm-4.7	Z.AI	open	85.9%
#42	Qwen3.5 27B qwen3.5-27b	Alibaba Cloud	open	85.8%
#43	Qwen3.5 122B qwen3.5-122b-a10b	Alibaba Cloud	open	85.7%
#44	GPT-5 gpt-5	OpenAI	closed	85.4%
#45	MiMo-V2.5 mimo-v2.5	Xiaomi	open	84.9%
#46	MiniMax M2.5 minimax-m2.5	MiniMax	open	84.8%
#47	Grok 4 Fast grok-4-fast	xAI	closed	84.7%
#48	MiMo-V2-Flash mimo-v2-flash	Xiaomi	open	84.6%
#49	o3-pro o3-pro	OpenAI	closed	84.5%
#50	Qwen3.5 35B qwen3.5-35b-a3b	Alibaba Cloud	open	84.5%
#51	Gemini 2.5 Pro gemini-2.5-pro	Google	closed	84.4%
#52	Gemma 4 31B gemma-4-31b	Google	open	84.3%
#53	Qwen3.6 27B qwen3.6-27b	Alibaba Cloud	open	84.2%
#54	DeepSeek V3.2 deepseek-v3.2	DeepSeek	open	84.0%
#55	Kimi K2 Thinking kimi-k2-thinking	Moonshot AI	open	83.8%
#56	GPT-5 Codex gpt-5-codex	OpenAI	closed	83.7%
#57	Claude Sonnet 4.5 claude-sonnet-4-5	Anthropic	closed	83.4%
#58	MiniMax M2.1 minimax-m2.1	MiniMax	open	83.0%
#59	GPT-5.4 nano gpt.5.4.nano	OpenAI	closed	82.8%
#60	o3 o3	OpenAI	closed	82.7%
#61	Qwen3.5 Omni Plus qwen3.5-omni-plus	Alibaba Cloud	closed	82.6%
#62	Claude Opus 4.1 claude-opus-4-1	Anthropic	closed	80.9%
#63	GPT-5 mini gpt-5-mini	OpenAI	closed	80.3%
#64	Nemotron 3 Super nemotron-3-super-120b	NVIDIA	open	80%
#65	Claude Opus 4 claude-opus-4	Anthropic	closed	79.6%
#66	Gemini 2.5 Flash gemini-2.5-flash	Google	closed	79.3%
#67	DeepSeek V3.1 Terminus deepseek-v3.1-terminus	DeepSeek	open	79.2%
#68	Gemma 4 26B gemma-4-26b-a4b	Google	open	79.2%
#69	Grok 3 mini grok-3-mini	xAI	closed	79.1%
#70	Nova 2 Pro nova-2-pro-preview	Amazon	closed	78.5%
#71	o4-mini o4-mini	OpenAI	closed	78.4%
#72	GLM-4.6 glm-4.6	Z.AI	open	78.0%
#73	Claude Sonnet 4 claude-sonnet-4	Anthropic	closed	77.7%
#74	MiniMax M2 minimax-m2	MiniMax	open	77.7%
#75	Claude Sonnet 3.7 claude-3-7-sonnet	Anthropic	closed	77.2%
#76	Mistral Small 4 mistral-small-2603	Mistral AI	open	76.9%
#77	Kimi K2 0905 kimi-k2-0905	Moonshot AI	open	76.7%
#78	Command A+ command-a-plus	Cohere	open	76.1%
#79	Gemma 4 12B gemma-4-12b	Google	open	75.3%
#80	Ling 2.6 1T ling-2.6-1t	InclusionAI	open	75.2%
#81	Mistral Medium 3.5 mistral-medium-3-5	Mistral AI	open	74.8%
#82	o1 o1	OpenAI	closed	74.7%
#83	Grok Code Fast 1 grok-code-fast-1	xAI	closed	72.7%
#84	Mistral Large 3 mistral-large-2512	Mistral AI	open	68.0%
#85	Claude Haiku 4.5 claude-haiku-4-5-20251001	Anthropic	closed	64.7%

The top of this snapshot is led by GPT-5.6 Sol at 94.6%; third place is 0.50 points behind. The top-10 spread is 1.8 points.

Compare top models

GPT-5.6 Sol vs Gemini 3.1 Pro Preview Gemini 3.1 Pro Preview vs Claude Mythos 5 Claude Mythos 5 vs Claude Opus 4.8 Claude Opus 4.8 vs GPT-5.5

About GPQA Diamond

Definition and scoring

Organisation: GPQA authors
Category: Reasoning
Version: Not evaluated
Direction: higher is better
Ranking use: Weighted 2.5%
Contamination risk: Unknown

The Diamond subset of a graduate-level, expert-written question-answering benchmark. Results stay tied to the exact model variant and evaluation system. Multiple systems for the same model use the best published score on this page; overall Lumina scoring uses the median of ranking-eligible rows.

Open benchmark source ↗

FAQ

Quick answers

What does GPQA Diamond measure?

The Diamond subset of a graduate-level, expert-written question-answering benchmark.

Which model scores highest on GPQA Diamond?

GPT-5.6 Sol by OpenAI currently leads with 94.6%.

How many models are evaluated on GPQA Diamond?

85 models in the LuminaBench cohort have a qualifying score on this benchmark.

Does this affect overall Lumina rank?

Yes. This family is ranking-weighted at 2.5% of its capability category. GPT-5.6 Sol is overall #4.

Humanity's Last Exam

Rolling · 105 results · Weighted 2.5%

ARC-AGI-2

2 · 10 results · Weighted 2.5%

LiveBench

Rolling · 17 results · Weighted 2.5%

ARC-AGI-3

3 · 3 results · Not Weighted

BenchLM Reasoning prior

bench-align-v5.1 · 18 results · Weighted 1.1%

SuperGPQA

Rolling · 8 results · Weighted 2.5%

Reasoning · Benchmark profile

GPQA Diamond

The Diamond subset of a graduate-level, expert-written question-answering benchmark.

Ranking weightedCurrent / rollingActive85 models

Data verified 15 July 2026 · Methodology 1.2.0

Benchmark score on GPQA Diamond

Leaderboard · 85 models

GPT-5.6 Sol leads at 94.6%, followed by Gemini 3.1 Pro Preview (94.1%) and Claude Mythos 5 (94.1%).

Ranking · 2.5%ReasoningCurrent

#1Closed

GPT-5.6 Sol

OpenAI

gpt-5.6-sol

94.6%

Overall #4 · Context 1.05M tokens

#2Closed

Gemini 3.1 Pro Preview

Google

gemini-3.1-pro-preview

94.1%

Overall #42 · Context 1.05M tokens

#3Closed

Claude Mythos 5

Anthropic

claude-mythos-5

94.1%

Overall #1 · Context 1M tokens

Visual analysis

See the field, not just the table.

Three views of the same sourced leaderboard: model placement, the score curve, and descriptive provider averages. These visuals never alter ranking calculations.

ModelsColours

OpenAIGoogleAnthropicMoonshot AIxAIMiniMaxAlibaba CloudZ.AI

Leaderboard bars

20 ranked models · higher is better

View accessible chart data

Accessible data for Leaderboard bars
Model	Rank	Provider	Score
GPT-5.6 Sol	#1	OpenAI	94.6%
Gemini 3.1 Pro Preview	#2	Google	94.1%
Claude Mythos 5	#3	Anthropic	94.1%
Claude Opus 4.8	#4	Anthropic	93.6%
GPT-5.5	#5	OpenAI	93.6%
Kimi K3	#6	Moonshot AI	93.5%
Grok 4.5	#7	xAI	93.1%
MiniMax M3	#8	MiniMax	92.9%
GPT-5.6 Terra	#9	OpenAI	92.9%
GPT-5.4	#10	OpenAI	92.8%
Claude Fable 5	#11	Anthropic	92.6%
Qwen3.7-Max	#12	Alibaba Cloud	92.4%
GPT-5.6 Luna	#13	OpenAI	92.3%
Gemini 3.5 Flash	#14	Google	92.2%
GPT-5.3-Codex	#15	OpenAI	91.5%
Claude Opus 4.6	#16	Anthropic	91.3%
GLM-5.2	#17	Z.AI	91.2%
Grok 4.20	#18	xAI	91.1%
Kimi K2.6	#19	Moonshot AI	91.1%
Claude Sonnet 5	#20	Anthropic	91.1%

Score curve

Shows how quickly scores separate down the field

View accessible chart data

Accessible data for Score curve
Model	Rank	Provider	Score
GPT-5.6 Sol	#1	OpenAI	94.6%
Gemini 3.1 Pro Preview	#2	Google	94.1%
Claude Mythos 5	#3	Anthropic	94.1%
Claude Opus 4.8	#4	Anthropic	93.6%
GPT-5.5	#5	OpenAI	93.6%
Kimi K3	#6	Moonshot AI	93.5%
Grok 4.5	#7	xAI	93.1%
MiniMax M3	#8	MiniMax	92.9%
GPT-5.6 Terra	#9	OpenAI	92.9%
GPT-5.4	#10	OpenAI	92.8%
Claude Fable 5	#11	Anthropic	92.6%
Qwen3.7-Max	#12	Alibaba Cloud	92.4%
GPT-5.6 Luna	#13	OpenAI	92.3%
Gemini 3.5 Flash	#14	Google	92.2%
GPT-5.3-Codex	#15	OpenAI	91.5%
Claude Opus 4.6	#16	Anthropic	91.3%
GLM-5.2	#17	Z.AI	91.2%
Grok 4.20	#18	xAI	91.1%
Kimi K2.6	#19	Moonshot AI	91.1%
Claude Sonnet 5	#20	Anthropic	91.1%

Provider profile

Average within the visible cohort · descriptive only

View accessible chart data

Accessible data for Provider profile
Provider	Models	Average score
Google	2	93.2%
OpenAI	6	93.0%
MiniMax	1	92.9%
Anthropic	5	92.5%
Alibaba Cloud	1	92.4%
Moonshot AI	2	92.3%
xAI	2	92.1%
Z.AI	1	91.2%

Benchmark score table (85 models)

One best score per model · higher is better

GPQA Diamond leaderboard ranked by score
Rank	Model	Provider	License	Score
#1	GPT-5.6 Sol gpt-5.6-sol	OpenAI	closed	94.6%
#2	Gemini 3.1 Pro Preview gemini-3.1-pro-preview	Google	closed	94.1%
#3	Claude Mythos 5 claude-mythos-5	Anthropic	closed	94.1%
#4	Claude Opus 4.8 claude-opus-4-8	Anthropic	closed	93.6%
#5	GPT-5.5 gpt-5.5	OpenAI	closed	93.6%
#6	Kimi K3 kimi-k3	Moonshot AI	closed	93.5%
#7	Grok 4.5 grok-4.5	xAI	closed	93.1%
#8	MiniMax M3 MiniMax-M3	MiniMax	—	92.9%
#9	GPT-5.6 Terra gpt-5.6-terra	OpenAI	closed	92.9%
#10	GPT-5.4 gpt-5.4	OpenAI	closed	92.8%
#11	Claude Fable 5 claude-fable-5	Anthropic	closed	92.6%
#12	Qwen3.7-Max qwen3.7-max	Alibaba Cloud	—	92.4%
#13	GPT-5.6 Luna gpt-5.6-luna	OpenAI	closed	92.3%
#14	Gemini 3.5 Flash gemini-3.5-flash	Google	closed	92.2%
#15	GPT-5.3-Codex gpt-5.3-codex	OpenAI	closed	91.5%
#16	Claude Opus 4.6 claude-opus-4-6	Anthropic	closed	91.3%
#17	GLM-5.2 glm-5-2	Z.AI	open	91.2%
#18	Grok 4.20 grok-4.20	xAI	closed	91.1%
#19	Kimi K2.6 kimi-k2.6	Moonshot AI	open	91.1%
#20	Claude Sonnet 5 claude-sonnet-5	Anthropic	closed	91.1%
#21	Hy3 hy3	Tencent	open	90.4%
#22	Qwen3.6 Plus qwen3-6-plus	Alibaba Cloud	closed	90.4%
#23	GPT-5.2 gpt-5.2	OpenAI	closed	90.3%
#24	Qwen3.7-Plus qwen3.7-plus	Alibaba Cloud	—	90.3%
#25	Grok 4.3 grok-4.3	xAI	closed	90.1%
#26	Claude Sonnet 4.6 claude-sonnet-4-6	Anthropic	closed	89.9%
#27	GPT-5.2 Codex gpt-5.2-codex	OpenAI	closed	89.9%
#28	Muse Spark 1.1 muse-spark-1-1	Meta	closed	89.8%
#29	Kimi K2.7 Code kimi-k2.7-code	Moonshot AI	open	89.6%
#30	DeepSeek V4 Flash deepseek-v4-flash	DeepSeek	open	89.4%
#31	Qwen3.5 397B qwen3.5-397b-a17b	Alibaba Cloud	open	89.3%
#32	DeepSeek V4 Pro deepseek-v4-pro	DeepSeek	open	88.8%
#33	Qwen3.6 Max qwen3.6-max	Alibaba Cloud	closed	88.8%
#34	Grok 4 grok-4	xAI	closed	87.7%
#35	Kimi K2.5 kimi-k2.5	Moonshot AI	—	87.6%
#36	GPT-5.4 mini gpt-5.4-mini	OpenAI	closed	87.5%
#37	Claude Opus 4.5 claude-opus-4-5	Anthropic	closed	87%
#38	Nemotron 3 Ultra nemotron-3-ultra-550b	NVIDIA	open	86.7%
#39	Qwen3 Max qwen3-max	Alibaba Cloud	closed	86.1%
#40	GLM-5 glm-5	Z.AI	open	86%
#41	GLM-4.7 glm-4.7	Z.AI	open	85.9%
#42	Qwen3.5 27B qwen3.5-27b	Alibaba Cloud	open	85.8%
#43	Qwen3.5 122B qwen3.5-122b-a10b	Alibaba Cloud	open	85.7%
#44	GPT-5 gpt-5	OpenAI	closed	85.4%
#45	MiMo-V2.5 mimo-v2.5	Xiaomi	open	84.9%
#46	MiniMax M2.5 minimax-m2.5	MiniMax	open	84.8%
#47	Grok 4 Fast grok-4-fast	xAI	closed	84.7%
#48	MiMo-V2-Flash mimo-v2-flash	Xiaomi	open	84.6%
#49	o3-pro o3-pro	OpenAI	closed	84.5%
#50	Qwen3.5 35B qwen3.5-35b-a3b	Alibaba Cloud	open	84.5%
#51	Gemini 2.5 Pro gemini-2.5-pro	Google	closed	84.4%
#52	Gemma 4 31B gemma-4-31b	Google	open	84.3%
#53	Qwen3.6 27B qwen3.6-27b	Alibaba Cloud	open	84.2%
#54	DeepSeek V3.2 deepseek-v3.2	DeepSeek	open	84.0%
#55	Kimi K2 Thinking kimi-k2-thinking	Moonshot AI	open	83.8%
#56	GPT-5 Codex gpt-5-codex	OpenAI	closed	83.7%
#57	Claude Sonnet 4.5 claude-sonnet-4-5	Anthropic	closed	83.4%
#58	MiniMax M2.1 minimax-m2.1	MiniMax	open	83.0%
#59	GPT-5.4 nano gpt.5.4.nano	OpenAI	closed	82.8%
#60	o3 o3	OpenAI	closed	82.7%
#61	Qwen3.5 Omni Plus qwen3.5-omni-plus	Alibaba Cloud	closed	82.6%
#62	Claude Opus 4.1 claude-opus-4-1	Anthropic	closed	80.9%
#63	GPT-5 mini gpt-5-mini	OpenAI	closed	80.3%
#64	Nemotron 3 Super nemotron-3-super-120b	NVIDIA	open	80%
#65	Claude Opus 4 claude-opus-4	Anthropic	closed	79.6%
#66	Gemini 2.5 Flash gemini-2.5-flash	Google	closed	79.3%
#67	DeepSeek V3.1 Terminus deepseek-v3.1-terminus	DeepSeek	open	79.2%
#68	Gemma 4 26B gemma-4-26b-a4b	Google	open	79.2%
#69	Grok 3 mini grok-3-mini	xAI	closed	79.1%
#70	Nova 2 Pro nova-2-pro-preview	Amazon	closed	78.5%
#71	o4-mini o4-mini	OpenAI	closed	78.4%
#72	GLM-4.6 glm-4.6	Z.AI	open	78.0%
#73	Claude Sonnet 4 claude-sonnet-4	Anthropic	closed	77.7%
#74	MiniMax M2 minimax-m2	MiniMax	open	77.7%
#75	Claude Sonnet 3.7 claude-3-7-sonnet	Anthropic	closed	77.2%
#76	Mistral Small 4 mistral-small-2603	Mistral AI	open	76.9%
#77	Kimi K2 0905 kimi-k2-0905	Moonshot AI	open	76.7%
#78	Command A+ command-a-plus	Cohere	open	76.1%
#79	Gemma 4 12B gemma-4-12b	Google	open	75.3%
#80	Ling 2.6 1T ling-2.6-1t	InclusionAI	open	75.2%
#81	Mistral Medium 3.5 mistral-medium-3-5	Mistral AI	open	74.8%
#82	o1 o1	OpenAI	closed	74.7%
#83	Grok Code Fast 1 grok-code-fast-1	xAI	closed	72.7%
#84	Mistral Large 3 mistral-large-2512	Mistral AI	open	68.0%
#85	Claude Haiku 4.5 claude-haiku-4-5-20251001	Anthropic	closed	64.7%

The top of this snapshot is led by GPT-5.6 Sol at 94.6%; third place is 0.50 points behind. The top-10 spread is 1.8 points.

Compare top models

GPT-5.6 Sol vs Gemini 3.1 Pro Preview Gemini 3.1 Pro Preview vs Claude Mythos 5 Claude Mythos 5 vs Claude Opus 4.8 Claude Opus 4.8 vs GPT-5.5

About GPQA Diamond

Definition and scoring

Organisation: GPQA authors
Category: Reasoning
Version: Not evaluated
Direction: higher is better
Ranking use: Weighted 2.5%
Contamination risk: Unknown

Open benchmark source ↗

FAQ

Quick answers

What does GPQA Diamond measure?

The Diamond subset of a graduate-level, expert-written question-answering benchmark.

Which model scores highest on GPQA Diamond?

GPT-5.6 Sol by OpenAI currently leads with 94.6%.

How many models are evaluated on GPQA Diamond?

85 models in the LuminaBench cohort have a qualifying score on this benchmark.

Does this affect overall Lumina rank?

Yes. This family is ranking-weighted at 2.5% of its capability category. GPT-5.6 Sol is overall #4.

Humanity's Last Exam

Rolling · 105 results · Weighted 2.5%

ARC-AGI-2

2 · 10 results · Weighted 2.5%

LiveBench

Rolling · 17 results · Weighted 2.5%

ARC-AGI-3

3 · 3 results · Not Weighted

BenchLM Reasoning prior

bench-align-v5.1 · 18 results · Weighted 1.1%

SuperGPQA

Rolling · 8 results · Weighted 2.5%

Leaderboard · 85 models

GPT-5.6 Sol

Gemini 3.1 Pro Preview

Claude Mythos 5

See the field, not just the table.

Benchmark score table (85 models)

Compare top models

Definition and scoring

Quick answers

What does GPQA Diamond measure?

Which model scores highest on GPQA Diamond?

How many models are evaluated on GPQA Diamond?

Does this affect overall Lumina rank?

More Reasoning benchmarks

Leaderboard · 85 models

GPT-5.6 Sol

Gemini 3.1 Pro Preview

Claude Mythos 5

See the field, not just the table.

Benchmark score table (85 models)

Compare top models

Definition and scoring

Quick answers

What does GPQA Diamond measure?

Which model scores highest on GPQA Diamond?

How many models are evaluated on GPQA Diamond?

Does this affect overall Lumina rank?

More Reasoning benchmarks