Best open-weight AI models
Models whose weights are published, so you can run them on your own hardware or pick among many API providers. Ranked by the overall index from Epoch AI’s independent tests.
Our picks
The ranking
Compare the top two| # | Model | Overall indexi | Inputi | Outputi | Contexti |
|---|---|---|---|---|---|
| 1 | Kimi K3Moonshot AI | 157.7 | $3 | $15 | 1M |
| 2 | GLM 5.3Z.ai (Zhipu) | 155.6 | $1.40 | $4.40 | 1M |
| 3 | DeepSeek V4 Pro 0813DeepSeek | 155.4 | $0.66 | $1.98 | 1M |
| 4 | DeepSeek V4.1 FlashDeepSeek | 155.0 | $0.30 | $1.20 | 1M |
| 5 | DeepSeek V4 Flash 0731DeepSeek | 154.5 | $0.018 | $0.32 | 1M |
| 6 | GLM 5.3 FlashZ.ai (Zhipu) | 151.9 | $0.15 | $0.50 | 1M |
| 7 | GLM 5.2Z.ai (Zhipu) | 151.8 | $0.14 | $3.99 | 1M |
| 8 | Kimi K2.6Moonshot AI | 151.0 | $0.65 | $3.41 | 256K |
| 9 | Inkling SmallThinking Machines Lab | 150.1 | $0.45 | $1.20 | 512K |
| 10 | Kimi K2.7 CodeMoonshot AI | 150.0 | $0.66 | $3.30 | 256K |
per 1M tokens · USD
Quick answers
- What is the best open-source AI model?
- Kimi K3 from Moonshot AI leads this ranking (Overall index: 157.7). It costs $3 per million input tokens and $15 per million output tokens.
- Is there a cheaper alternative?
- DeepSeek V4 Flash 0731 from DeepSeek (Overall index: 154.5) costs $0.018 / $0.32 per million tokens, about 64× less than Kimi K3.
How we rank
Scores come from tests run independently by Epoch AI (CC BY 4.0), not from the companies’ own announcements. Prices come from OpenRouter’s public API. Both are refreshed automatically several times a day, so this ranking follows new releases and price changes.
Other rankings
CodingMathScienceCheapFreeLong contextImages & PDFsRankings →