BrowseComp

Name: BrowseComp
Keywords: BrowseComp, AI benchmark, model evaluation, leaderboard, AI performance

AI benchmark results and model performance

Report an issue

Type: percentageAgents

Recorded Results

Average Score

65.99%

Score Range

21.30% - 91.50%

Leading Model (lowest score)

21.30% - GLM 4.5 Air

Scores Over Time

Individual benchmark scores plotted by date.

Model Results

Organisation	Model	Reported	Top Score	Info	Self Reported	Source
Anthropic	Claude Mythos Preview	07 Apr 2026	86.90%	4.9x fewer tokens vs Opus 4.6	Yes	Source
Anthropic	Claude Opus 4.6	05 Feb 2026	N/A	-	Yes	-
z.AI	GLM 4.5 Air	28 Jul 2025	21.30%	-	Yes	Source
z.AI	GLM 4.5	28 Jul 2025	26.40%	-	Yes	Source
DeepSeek	DeepSeek V3.1	21 Aug 2025	30%	Search agent with commercial API + webpage filter + 128K context	Yes	Source
DeepSeek	DeepSeek V3.1 Terminus	22 Sept 2025	30%	inferred alias from deepseek-v3.1	Yes	Source
Nvidia	Nemotron 3 Super	11 Mar 2026	31.28%	-	Yes	Source
Nvidia	Nemotron 3.5 Lightning 30B A3B	11 Aug 2026	36.97%	NVIDIA release evaluation suite	Yes	Source
DeepSeek	DeepSeek OCR 2	-	40.10%	inferred family alias from deepseek-v3.2-exp (score=0.3809; benches=14)	Yes	Source
DeepSeek	DeepSeek V3.2 Exp	29 Sept 2025	40.10%	-	Yes	Source
z.AI	GLM 4.7 Flash	19 Jan 2026	42.80%	-	Yes	Source
MiniMax	MiniMax M2	27 Oct 2025	44%	-	Yes	-
MiniMax	MiniMax M2 Her	24 Jan 2026	44%	inferred modality/version alias from minimax-m2	Yes	-
z.AI	GLM 4.6	30 Sept 2025	45.10%	-	Yes	Source
Mistral	Mistral Medium 3.5	29 Apr 2026	48.60%	-	Yes	Source
OpenAI	o4 Mini	16 Apr 2025	51.50%	-	Yes	Source
OpenAI	o4 mini Deep Research	26 Jun 2025	51.50%	inferred modality/version alias from o4-mini	Yes	Source
z.AI	GLM 4.7	22 Dec 2025	52%	-	Yes	Source
OpenAI	GPT 5	07 Aug 2025	54.90%	With Thinking, Pass @ 1	Yes	Source
Xiaomi	MiMo V2 Flash	16 Dec 2025	58.30%	-	Yes	Source
Moonshot	Kimi K2 Thinking	06 Nov 2025	60.20%	inferred alias from kimi-k2-thinking-0905	Yes	Source
Qwen	Qwen 3.5 27B	24 Feb 2026	61%	-	Yes	Source
Qwen	Qwen 3.5 35B A3B	24 Feb 2026	61%	-	Yes	Source
Qwen	Qwen 3.5 Flash	23 Feb 2026	61%	inferred family alias from qwen3.5-27b (score=0.4147; benches=81)	Yes	Source
MiniMax	MiniMax M2.1	23 Dec 2025	62%	-	Yes	Source