Leaderboard
OSWorld-Verified
16 models scored · metric: success rate. Prices are the cheapest listed across all serving providers. Pts/$ = score ÷ blended price (3×input + output) / 4 — how much score each dollar buys.
Best value · top 5 by pts per dollar
- 1.MiniMax-M3133
- 2.Muse Glimmer 30B132
- 3.GPT-5.4 nano95.1
- 4.Gemini 3.5 Flash Lite87.1
- 5.Gemini 3.6 Flash55.3
| # | Model | Score | Best in /M | Best out /M | Pts / $ | Context |
|---|---|---|---|---|---|---|
| 🥇 | Claude Fable 5anthropic | 85 | $10 | $50 | 4.3 | 1M |
| 🥈 | Claude Opus 4.8anthropic | 83.4 | $5 | $25 | 8.3 | 1M |
| 🥉 | Gemini 3.6 Flashgoogle | 83 | $0.75 | $3.75 | 55.3 | 1.05M |
| 4 | Claude Sonnet 5anthropic | 81.2 | $2 | $10 | 20.3 | 1M |
| 5 | Muse Spark 1.1meta | 80.8 | $1.25 | $4.25 | 40.4 | 1M |
| 6 | GPT-5.5openai | 78.7 | $4.55 | $27.27 | 7.7 | 1.05M |
| 7 | Claude Sonnet 4.6anthropic | 78.5 | $3 | $15 | 13.1 | 1M |
| 8 | Gemini 3.5 Flashgoogle | 78.4 | $1.50 | $9 | 23.2 | 1.05M |
| 9 | Claude Opus 4.7anthropic | 78 | $5 | $25 | 7.8 | 1M |
| 10 | Gemini 3.1 Pro Previewgoogle | 76.2 | $2 | $12 | 16.9 | 1.05M |
| 11 | GPT-5.4openai | 75 | $2.20 | $14 | 14.6 | 1.05M |
| 12 | Gemini 3.5 Flash Litegoogle | 74 | $0.3 | $2.50 | 87.1 | 1.05M |
| 13 | GPT-5.4 miniopenai | 72.1 | $0.68 | $4 | 47.7 | 400K |
| 14 | MiniMax-M3minimax | 70.06 | $0.3 | $1.20 | 133 | 512K |
| 15 | Muse Glimmer 30Bmeta | 65.9 | $0.3 | $1.10 | 132 | 131K |
| 16 | GPT-5.4 nanoopenai | 39 | $0.18 | $1.10 | 95.1 | 400K |
Cheapest scorer: GPT-5.4 nano at $0.18 input /M.