Best open-weight models in 2026
The strongest open-weight models in the index, ranked by average benchmark score. 371 models fall into this group.
What the data shows
DeepSeek-V3.2 leads the open-weight group at 84.0. The group median is 67.4.
The price range runs from $0.040 to $5.91 per million input tokens, so 'best' and 'cheapest' are different shortlists.
How to read the ranking
Open-weight models can be self-hosted, which changes the cost model and the data-residency story. The listed API price is only one deployment option.
- Check the licence before planning a deployment.
- Compare hosted and self-hosted cost if open weights are available.
- Test on your own data.
- Keep a fallback from the other licensing group.
Best open-weight models
| # | Model | Provider | Input / 1M | Output / 1M | Context | Avg score |
|---|---|---|---|---|---|---|
| 1 | DeepSeek-V3.2 | DeepSeek | $5.88 | $24.70 | 1,000,000 | 84.0 |
| 2 | DeepSeek-V3.2 Instruct | DeepSeek | $5.91 | $29.01 | 128,000 | 83.7 |
| 3 | DeepSeek-R1 Instruct | DeepSeek | $5.90 | $26.49 | 1,000,000 | 83.1 |
| 4 | Kimi K2 | Moonshot AI | $5.85 | $21.72 | 128,000 | 83.0 |
| 5 | DeepSeek-R1 | DeepSeek | $5.83 | $25.23 | 128,000 | 82.8 |
| 6 | DeepSeek-V3.2 Chat | DeepSeek | $5.80 | $26.43 | 1,000,000 | 82.5 |
| 7 | DeepSeek-R1 Chat | DeepSeek | $5.62 | $21.49 | 400,000 | 82.4 |
| 8 | Kimi K2 Instruct | Moonshot AI | $5.80 | $23.99 | 1,000,000 | 82.3 |
| 9 | Kimi K2 Chat | Moonshot AI | $5.74 | $22.24 | 400,000 | 82.1 |
| 10 | Llama 4 Maverick Instruct | Meta AI | $5.33 | $23.04 | 200,000 | 81.7 |
| 11 | Mistral Large 3 | Mistral AI | $5.41 | $23.83 | 128,000 | 81.7 |
| 12 | Llama 4 Maverick | Meta AI | $5.36 | $25.19 | 200,000 | 81.6 |
| 13 | Mistral Large 3 Instruct | Mistral AI | $5.31 | $22.47 | 400,000 | 81.1 |
| 14 | Kimi K1.5 | Moonshot AI | $5.19 | $21.20 | 128,000 | 80.7 |
| 15 | MiniMax-M2 | MiniMax | $5.19 | $22.73 | 200,000 | 80.6 |
| 16 | Kimi K1.5 Instruct | Moonshot AI | $5.15 | $28.20 | 200,000 | 80.5 |
| 17 | GLM-4.6 Instruct | Zhipu AI | $5.15 | $24.32 | 1,000,000 | 80.4 |
| 18 | Qwen3 | Alibaba Cloud | $5.04 | $22.52 | 128,000 | 80.3 |
| 19 | GLM-4.6 | Zhipu AI | $5.10 | $18.45 | 128,000 | 80.2 |
| 20 | MiniMax-M2 Chat | MiniMax | $5.02 | $18.95 | 200,000 | 80.2 |
Methodology
This story uses the AI & LLM Index dataset: 600 models, 40 providers and 12 benchmarks. Prices are listed per million tokens. Scores are modelled estimates on a 0–100 scale and are not official benchmark results. Figures are rounded for readability.
Related
FAQ
What does the Best open-weight models in 2026 data show?
The strongest open-weight models in the index, ranked by average benchmark score. 371 models fall into this group.
Are the numbers official?
No. Prices and scores are compiled for this demo index. Benchmark values are modelled estimates, not official results. Verify figures with the provider or benchmark owner before making decisions.
How should I use this ranking?
Use it to form a shortlist. Check the specific model pages, compare price and context, and test the shortlist on your own examples before committing.