What does Artificial Analysis · GPQA Diamond measure?

Artificial Analysis · GPQA Diamond is a knowledge benchmark in the BenchGecko catalog. 55 AI models have been tested on it. Scores range from 11.1 to 96.1 out of 100.

Which model leads on Artificial Analysis · GPQA Diamond?

GPT-6 Astra from OpenAI leads Artificial Analysis · GPQA Diamond with a score of 96.1. The median score across 55 tested models is 79.2.

Is Artificial Analysis · GPQA Diamond saturated?

Yes · the top model on Artificial Analysis · GPQA Diamond has reached 96.1 out of 100, within 5% of the theoretical ceiling. This benchmark is approaching saturation and may be replaced by a harder successor.

Does Artificial Analysis · GPQA Diamond predict performance on other benchmarks?

Yes · Artificial Analysis · GPQA Diamond scores correlate 0.98 with ARC-AGI across 9 shared models. Models that do well on Artificial Analysis · GPQA Diamond tend to do well on ARC-AGI.

How often is Artificial Analysis · GPQA Diamond data refreshed?

BenchGecko pulls updates daily. New model scores on Artificial Analysis · GPQA Diamond appear as soon as they are published by Epoch AI or the model provider.

Benchmark · KnowledgeSettled

Artificial Analysis · GPQA Diamond

Name: Artificial Analysis · GPQA Diamond Benchmark
Creator: BenchGecko
License: https://creativecommons.org/licenses/by/4.0/

Updated 2026-09-22

Models tested

Top score

96.1

GPT-6 Astra

Median

79.2

min 11.1

Top-5 spread

σ 1.0

Settled

The Frontier

Best score over time · one chart, every benchmark

Chart type

Frontier on Artificial Analysis · GPQA Diamond rose from 57.5 to 96.1 in 20 months · +38.6 points · latest leader GPT-6 Astra from OpenAI.

Pink dots = frontier records · 8 totalClick to open model page

Full rankings

55 models tested · sorted by score

#	Model	Score	Price
1	GPT-6 Astra· OpenAI	96.1	$10.00
2	Gemini 3.8 Flash· Google DeepMind	95.3	$0.75
3	Gemini 3.1 Pro Preview· Google DeepMind	94.1	$2.00
4	Claude Fable 5.1· Anthropic	93.7	$10.00
5	Kimi K3· moonshotai	93.5	$2.70
6	Qwen3.8 2.4T A95B· Alibaba Qwen	93.5	$2.00
7	MiniMax M3· minimax	92.9	$0.30
8	DeepSeek V4 Pro· DeepSeek	92.8	$0.21
9	Qwen3.8 Max (0902)· Alibaba Qwen	92.8	$2.00
10	GPT-5.6 Terra· OpenAI	92.5	$2.00
11	GLM 5.3· z-ai	91.7	$1.40
12	GPT-5.3-Codex· OpenAI	91.5	$1.75
13	GLM 5.3 Flash· z-ai	91.2	$0.15
14	Qwen3.8 27B· Alibaba Qwen	90.5	$0.42
15	Qwen3.7 Plus· Alibaba Qwen	90.0	$0.32
16	Hy3· tencent	89.7	$0.13
17	Kimi K2.7 Code· moonshotai	89.6	$0.61
18	Solar Pro 4· upstage	89.1	$0.09
19	Nemotron 3 Ultra· NVIDIA	86.7	$0.50
20	Qwen3.5 397B A17B· Alibaba Qwen	86.1	$0.55
21	Gemma 4 31B· Google DeepMind	85.7	$0.09
22	N Nanbeige4.1 3B· Nanbeige	84.9	—
23	Qwen3.6 35B A3B· Alibaba Qwen	84.1	$0.15
24	Gemini 3.5 Flash Lite· Google DeepMind	83.8	$0.30
25	o3· OpenAI	82.7	$2.00
26	Qwen3.5-122B-A10B· Alibaba Qwen	82.7	$0.26
27	Nemotron 3 Super· NVIDIA	80.0	$0.08
28	Gemma 4 26B A4B · Google DeepMind	79.2	$0.07
29	Qwen3.5-9B· Alibaba Qwen	78.6	$0.10
30	gpt-oss-120b· OpenAI	78.2	$0.04
31	LongCat 2.0· meituan	78.0	$0.30
32	Qwen3.5 4B· Alibaba	77.1	—
33	Mistral Small 4· Mistral AI	76.9	$0.15
34	Command A+· Cohere	76.1	$0.30
35	Qwen3 Next 80B A3B Instruct· Alibaba Qwen	75.9	$0.09
36	North Mini Code (free)· Cohere	75.7	$0.00
37	Trinity Large Thinking· arcee-ai	75.2	$0.25
38	Mistral Medium 3.5· Mistral AI	74.8	$1.50
39	Nemotron 3.5 Lightning· NVIDIA	74.3	$0.06
40	Qwen3 Coder Next· Alibaba Qwen	73.7	$0.12
41	gpt-oss-20b· OpenAI	68.8	$0.02
42	Llama 4 Maverick· Meta	67.1	$0.19
43	Granite 4.2 8B· ibm-granite	64.4	$0.06
44	Llama 4 Scout· Meta	58.7	$0.10
45	Phi 4· Microsoft	57.5	$0.07
46	NVIDIA Nemotron Nano 9B V2· NVIDIA	57.0	—
47	LFM2.5-2.6B (free)· liquid	55.8	$0.00
48	Reka Flash 3· rekaai	52.9	$0.10
49	Command A· Cohere	52.7	$2.50
50	Qwen3.5 2B· Alibaba	45.6	—
51	Molmo2 8B· allenai	42.5	—
52	Granite 4.0 Micro· ibm-granite	33.6	$0.02
53	Phi 4 Mini Instruct· Microsoft	33.1	—
54	Phi 4 Multimodal Instruct· Microsoft	31.5	—
55	Qwen3.5 0.8B· Alibaba	11.1	—