Nebius Token Factory: Models and Prices
Inference provider serving open-weight models.
Nebius Token Factory served 328B tokens of other labs' open-weight models over Nov 2 to Nov 8, 2026 through a large public LLM router, #33 among providers.
Open-weight models
| Model | Family | Share of model | Tokens, 7d | Input $/M | Output $/M | Tokens/s | Uptime 3d | Context |
|---|---|---|---|---|---|---|---|---|
| Nemotron 3.5 Super VL | Nemotron | 2.8% | 104B | $0.30 | $0.90 | 77 | 99.5% | 1.05M |
| MiniMax M3.1 Flash | MiniMax | 0.9% | 46B | $0.15 | $0.60 | 103 | 100.0% | 1.05M |
| DeepSeek V4.1 Flash 1017 | DeepSeek | <0.1% | 38B | $0.30 | $1.20 | 96 | 100.0% | 1.05M |
| Qwen3.8 27B | Qwen | 5.9% | 35B | $0.45 | $3.00 | 250 | 99.2% | 1M |
| DeepSeek V4.1 Flash | DeepSeek | 0.1% | 20B | $0.30 | $1.20 | 71 | 99.9% | 1.05M |
| Beam 1.0 501B A23B | Other open-weight | 20.0% | 18B | $0.25 | $1.00 | 194 | 98.2% | 1.05M |
| Kimi K3.1 Pro | Kimi | 0.4% | 17B | $2.00 | $10.00 | 101 | 96.9% | 1.05M |
| Qwen3 30B A3B Instruct 2507 | Qwen | 100.0% | 16B | $0.10 | $0.30 | 51 | 96.1% | 262K |
| DeepSeek V4 Flash 0731 | DeepSeek | 0.2% | 7B | $0.14 | $0.28 | 27 | 96.7% | 1.05M |
| Muse Spark Open | Other open-weight | 5.2% | 7B | $1.25 | $4.25 | 146 | 100.0% | 1.05M |
| DeepSeek V4 Pro 0813 | DeepSeek | 1.0% | 6B | $1.32 | $3.96 | 59 | 99.8% | 1.05M |
| gpt-oss-120b | gpt-oss | 2.3% | 6B | $0.15 | $0.60 | 251 | 100.0% | 131K |
| GLM 5.3 Flash | GLM | <0.1% | 4B | $0.15 | $0.50 | 78 | 99.8% | 1.05M |
| Nemotron 3 Nano 30B A3B | Nemotron | 100.0% | 3B | $0.06 | $0.24 | 90 | 100.0% | 262K |
| GLM 5.1 | GLM | 7.4% | 959M | $1.40 | $4.40 | 35 | 99.4% | 203K |
| GLM 5.3 | GLM | - | - | $1.40 | $4.40 | 47 | 97.9% | 1.05M |
| GLM 5.2 | GLM | - | - | $1.40 | $4.40 | 163 | 81.0% | 1.05M |
| DeepSeek V4 Pro 0423 | DeepSeek | - | - | $1.75 | $3.50 | 40 | 97.9% | 1.05M |
| Gemma 3 27B | Gemma | - | - | $0.10 | $0.30 | 34 | 98.4% | 110K |
| Hermes 4 405B | Other open-weight | - | - | $1.00 | $3.00 | 41 | 100.0% | 131K |
List prices per million tokens as of November 8, 2026. Share: of all tokens served on that model; a dash where the volume is not reported.
Company
- Website
- docs.nebius.com
- Headquarters
- Netherlands
Finance your GPUs
Serving tokens on your own GPUs? Tell us the cluster, the models you serve and the contracts behind them, and we will share it with partner funders that fit.
Prefer email? hello@amcompute.com