Open Source Models
Open-weight models you can self-host or fine-tune, from providers like Meta, DeepSeek, Mistral, and Qwen.
Based on 10 AI models
| Model | Provider | Score | Price (in) | Context |
|---|---|---|---|---|
| Mistral Large 2 | Mistral | 73.6 | $2.00 | 128 000 |
| Llama 3.3 70B | Meta | 63 | $0.59 | 128 000 |
| Qwen2.5 72B | Alibaba | 62.6 | $1.20 | 131 072 |
| DeepSeek V3 | DeepSeek | 59.4 | $0.27 | 65 536 |
| Llama 3.1 405B | Meta | 58.1 | $3.00 | 128 000 |
| DeepSeek R1 | DeepSeek | 55 | $0.55 | 65 536 |
| Llama 4 Maverick | Meta | 44.8 | $0.25 | 1 000 000 |
| Llama 4 Scout | Meta | 44.8 | $0.10 | 10 000 000 |
| Qwen3 Coder 480B | Alibaba | 20.7 | $0.30 | 262 144 |
| DeepSeek V4-Flash | DeepSeek | 20.7 | $0.14 | 1 048 576 |
Frequently Asked Questions
What is the best model for Open Source Models?
Based on our data, Mistral Large 2 ranks first for Open Source Models with a score of 74.
How were these models selected?
Models are ranked by scenario-specific data: benchmark scores (coding/reasoning/math/vision), capability support, context window, or price. All data comes from our transparent model catalog, updated daily.
How fresh is this data?
Model data, prices, and rankings are refreshed on a regular schedule; ranking scores are snapshotted daily. Each model page shows its last verified date.