GPQA Diamond — cost vs performance

Best current row per canonical model in the group “accuracy · variant=GPQA Diamond · evaluator=Artificial Analysis” (459 models) against context window (tokens). The dashed line is the Pareto frontier: no model is both better and cheaper than a point on it.

← Leaderboard

XOutput price Input price Parameters Context Memory (est.)Scalelog linearBubblecontext params none

open / restricted weights closed Pareto frontier (15 models)bubble = context window

Frontier models 15

Pareto frontier
Model	accuracy on GPQA Diamond	Rank	context window (tokens)	Provider	Trust
gpt-6-astraOpenAI · Closed	96.3%	1	1.05M	—	Independent
Gemini 3.8 FlashGoogle · Closed	95.3%	2	1.05M	—	Independent
Grok 4.6xAI · Closed	95.0%	3	500K	—	Independent
gpt-5.3-codexOpenAI · Closed	91.5%	27	400K	—	Independent
Kimi K2.6Moonshot AI · Open weights	91.1%	31	262.1K	—	Independent
grok-build-0-1-06-16SpaceXAI · Closed	89.5%	50	256K	—	Independent
MiniMax M2.7MiniMax · Open weights	87.4%	67	204.8K	—	Independent
deepseek-v3-2-specialeDeepSeek · Open weights	87.1%	71	128K	—	Independent
qwen3-omni-30b-a3b-instructAlibaba Group · Open weights	72.6%	208	65.5K	—	Independent
qwen3-235b-a22b-instructAlibaba Group · Open weights	70%	227	32.8K	—	Independent
tri-21b-think-v0-5Trillion Labs · Open weights	60.1%	277	32K	—	Independent
Phi 4Microsoft · Open weights	57.5%	295	16.4K	—	Independent
R1 Distill Llama 70BDeepSeek · Open weights	40.2%	373	8.19K	—	Independent
gpt-4OpenAI · Closed	35.0%	391	8.19K	—	Independent
olmo-2-32bAllen Institute for AI · Open weights	32.8%	405	4.1K	—	Independent

Methodology. Points are the best current row per canonical model in comparability group 'accuracy · variant=GPQA Diamond · evaluator=Artificial Analysis'. Price = cheapest current offer across providers (the provider shown). Pareto frontier maximises the score and minimises x; exact ties are all kept. memory_estimate is an estimate (see /methodology). Only the selected comparability group is plotted; points under other configurations are not mixed in. Price = cheapest current offer across providers at the time of the last crawl. Nothing is estimated except where marked.