Nebius Token Factory by Nebius
OpenAI-compatible inference API for open-weight LLMs, embeddings and image models on Nebius GPUs.
Model hosting and inference Official MCP server
＋ Build a stack with Nebius Token Factory Website Official pricing
Price
from $0.01/M
pay as you go, list price
Free tier
No
SDK downloads
—
no official SDK package
Pricing
| Free tier | — |
|---|---|
| Cheapest paid plan | — |
| Usage price | $0.15/M input, $0.6/M output (gpt-oss-120b); $0.3/M input, $1.2/M output (DeepSeek V4.1 Flash); $0.01/M (Qwen3-Embedding-8B) |
| Good to know | Formerly Nebius AI Studio. Signup includes free credits (amount not stated). The official MCP server covers Token Factory Sandboxes, not model inference. |
From the official pricing page, checked Oct 10, 2026. Prices change: confirm before you buy.
Use it from an agent
| MCP server | Official: https://docs.tokenfactory.nebius.com/sandboxes/mcp/index.md |
|---|
Alternatives to Nebius Token Factory
| API | Free | Price | SDK downloads / wk | GitHub stars | MCP |
|---|---|---|---|---|---|
| Fireworks Serverless inference API with per-token pricing for running open and proprietary models. |
— | per token for inference; per GPU second for on-demand deployments | — | — | |
| Replicate Run and deploy machine learning models with pay-per-use pricing for inference. |
— | Varies by model and hardware; examples: $0.04/output image,... | — | — | |
| AI21 API access to foundation models for text generation and chat tasks. |
$10 credits for 7 days | Free plan then $0.2 per 1M input tokens, $0.4 per 1M output tokens... |
— | — | |
| Baseten Deploy and run custom, fine-tuned, and open-source AI models with GPU compute. |
Free credits for new accounts | Free plan then Price per 1M tokens for Model APIs; price per minute for... |
— | — | |
| DeepInfra API for running various AI models including text generation, embeddings, image generation, and speech recognition. |
— | Per token pricing varies by model; e.g., $0.06 per 1M input tokens... | — | — | |
| Together AI API for inference on open-source and third-party language models, vision, audio, video, and embeddings. |
— | varies by model and task; e.g., $0.30 per 1M input tokens (MiniMax... | — | — |