Leaderboard

OSWorld-Verifiedofficial site ↗

17 models scored · metric: success rate. Prices are the cheapest listed across all serving providers. Cost/run and Pts/$ are priced for the agentic token profile — tool loops re-send a growing transcript and think between calls.

A benchmark is priced for the workload it exercises, not a fixed input:output blend: an agent loop and a single hard question bill very differently on the same rate card. Profile: 12k in · 2k out · 3k reasoning · 70% cache hit · 32k context.

Scores are facts as publicly reported by labs (see report links below and on each model page) and aggregated via models.dev. OSWorld-Verified is maintained by its own project — Model Pulse is not affiliated with or endorsed by it.

Best value · top 5 by pts per dollar · agentic profile

#
🥇	Qwen3.8 Max Previewalibaba	86.1	$2	$6	$0.043	2,007	1M
🥈	Claude Fable 5anthropic	85	$3	$18.50	$0.111	764	1M
🥉	Claude Opus 4.8anthropic	83.4	$0.425	$2.13	$0.016	5,058	1.05M
4	Gemini 3.6 Flashgoogle	83	$0.375	$1.88	$0.011	7,493	1.05M
5	Claude Sonnet 5anthropic	81.2	$1.44	$7.20	$0.056	1,453	1M
6	Muse Spark 1.1meta	80.8	$1.25	$4.25	$0.029	2,761	1.05M
7	GPT-5.5openai	78.7	$0.188	$1.13	0.821¢	9,583	1.05M
8	Claude Sonnet 4.6anthropic	78.5	$0.9	$5.55	$0.033	2,353	1M
9	Gemini 3.5 Flashgoogle	78.4	$0.186	$1.11	0.813¢	9,639	1.05M
10	Claude Opus 4.7anthropic	78	$4	$20	$0.067	1,168	1.05M
11	Gemini 3.1 Pro Previewgoogle	76.2	$1	$6	$0.035	2,191	1.05M
12	GPT-5.4openai	75	$0.75	$6	$0.035	2,163	1.05M
13	Gemini 3.5 Flash Litegoogle	74	$0.15	$1.25	0.699¢	10,585	1.05M
14	GPT-5.4 miniopenai	72.1	$0.375	$4	$0.022	3,227	400K
15	MiniMax-M3minimax	70.06	$0.225	$0.9	0.609¢	11,498	1.05M
16	Muse Glimmer 30Bmeta	65.9	$0.2	$0.8	0.550¢	11,982	131K
17	GPT-5.4 nanoopenai	39	$0.18	$1.10	0.662¢	5,888	1.05M

Score reports: alibabacloud.com · anthropic.com · deepmind.google · anthropic.com · ai.meta.com · openai.com — full source URLs are linked on each model page.

Cheapest scorer: Gemini 3.5 Flash Lite at $0.15 input /M.