# Nebius Token Factory API pricing and models

Nebius Token Factory publishes per-token prices for 27 text models, as of October 20, 2026. The lowest output rate is $0.24 per million tokens, on Nemotron-3-Nano-30B-A3B, and the highest-scoring model it serves is Kimi K3, at 79 on our 100-point score.

Token Factory, formerly Nebius AI Studio, is Nebius’s per-token API for open-weight models. Nebius’s GPU cloud is priced separately and is not covered here.

Source: [Nebius Token Factory prices (page JS-rendered; figures read from the host's public models_info JSON, which matches the catalog at tokenfactory.nebius.com/models)](https://nebius.com/token-factory/prices), read 2026-10-20. Prices in US dollars per million tokens.

## Models we score

| Model | Score | Input | Output | Cached input | Context | Listed speed | vs maker |
|---|---|---|---|---|---|---|---|
| Kimi-K3 (`moonshotai/Kimi-K3`) | 79 | $3.00 | $15.00 | — | 1.02M | 264.55 tok/s | Same |
| DeepSeek V4.1 Flash (`deepseek-ai/DeepSeek-V4.1-Flash`) | 73 | $0.30 | $1.20 | — | 1.05M | 139 tok/s | Same |
| Qwen3.8-27B (`Qwen/Qwen3.8-27B`) | 73 | $0.45 | $3.00 | — | 262K | 218 tok/s | — |
| Muse Spark Open (`meta-models/muse-spark-open`) | 72 | $1.25 | $4.25 | — | 1.05M | 143 tok/s | — |
| GLM-5.3-Flash (`zai-org/GLM-5.3-Flash`) | 69 | $0.15 | $0.50 | — | 1.02M | 349 tok/s | — |
| DeepSeek-V4-Pro-0813 (`deepseek-ai/DeepSeek-V4-Pro-0813`) | 69 | $1.32 | $3.96 | — | 979K | 37 tok/s | Same |
| Reflection Beam 1.0 (`reflection/beam`) | 68 | $0.25 | $1.00 | — | 1.05M | 389 tok/s | — |
| GLM-5.3 (`zai-org/GLM-5.3`) | 68 | $1.40 | $4.40 | — | 1.02M | 455 tok/s | — |
| Kimi-K2.6 (`moonshotai/Kimi-K2.6`) | 67 | $0.95 | $4.00 | — | 262K | 60 tok/s | Same |
| MiniMax-M3 (`MiniMaxAI/MiniMax-M3`) | 65 | $0.30 | $1.20 | — | 1.05M | 248 tok/s | Same |
| GLM-5.2 (`zai-org/GLM-5.2`) | 65 | $1.40 | $4.40 | — | 1.05M | 25 tok/s | Same |
| GLM-5.1 (`zai-org/GLM-5.1`) | 64 | $1.40 | $4.40 | — | 203K | 25 tok/s | Same |
| Kimi-K2.7-Code (`moonshotai/Kimi-K2.7-Code`) | 63 | $0.95 | $4.00 | — | 262K | 231.26 tok/s | Same |
| DeepSeek-V4-Flash-0731 (`deepseek-ai/DeepSeek-V4-Flash-0731`) | 62 | $0.14 | $0.28 | — | 1.02M | 351 tok/s | Same |
| gpt-oss-120b (`openai/gpt-oss-120b`) | 55 | $0.15 | $0.60 | — | 131K | 40 tok/s | — |
| Gemma-3-27b-it (`google/gemma-3-27b-it`) | 41 | $0.10 | $0.30 | — | 110K | 20 tok/s | — |
| Nemotron-3-Super-120b-a12b (`nvidia/nemotron-3-super-120b-a12b`) | 0 | $0.30 | $0.90 | — | 262K | 127 tok/s | — |

## Other models

| Model | Input | Output | Cached input | Context | Listed speed |
|---|---|---|---|---|---|
| Nemotron-3-Nano-30B-A3B (`nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B`) | $0.06 | $0.24 | — | 262K | 60 tok/s |
| Nemotron-3.5-Lightning (`nvidia/Nemotron-3_5-Lightning`) | $0.06 | $0.24 | — | 1.05M | 314.12 tok/s |
| Qwen3-30B-A3B-Instruct-2507 (`Qwen/Qwen3-30B-A3B-Instruct-2507`) | $0.10 | $0.30 | — | 262K | 70 tok/s |
| Qwen3-235B-A22B-Instruct-2507 (`Qwen/Qwen3-235B-A22B-Instruct-2507`) | $0.20 | $0.60 | — | 262K | 27 tok/s |
| Nemotron 3.5 Super VL (`nvidia/Nemotron-3.5-Super-VL`) | $0.30 | $0.90 | — | 1.05M | 282.4 tok/s |
| openbmb/MiniCPM-V-4_5 (`openbmb/MiniCPM-V-4_5`) | $0.66 | $1.11 | — | 32K | 49.5 tok/s |
| Hermes-4-405B (`NousResearch/Hermes-4-405B`) | $1.00 | $3.00 | — | 131K | 20 tok/s |
| Nemotron-3-Ultra-550b-a55b (`nvidia/Nemotron-3-Ultra-550b-a55b`) | $1.00 | $3.00 | — | 1.05M | 523 tok/s |
| DeepSeek-V4-Pro (`deepseek-ai/DeepSeek-V4-Pro`) | $1.75 | $3.50 | — | 1.05M | 24 tok/s |
| Qwen3.5-397B-A17B (`Qwen/Qwen3.5-397B-A17B`) | $0.60 | $3.60 | — | 262K | 80 tok/s |

## Billing notes

- Nebius describes its catalog prices as approximate and dependent on actual usage.
- Its docs describe a faster flavor of some models at a different price. The public catalog lists one flavor per model, which is what the table shows.

## Hosts that sell the same models

- [OpenRouter](https://benchlm.ai/hosts/openrouter): 17 shared, 11 cheaper per output token (DeepSeek V4.1 Flash 1017: $0.12 vs $1.20)
- [Novita AI](https://benchlm.ai/hosts/novita): 14 shared, 3 cheaper per output token (GPT-OSS 120B: $0.25 vs $0.60)
- [DeepInfra](https://benchlm.ai/hosts/deepinfra): 12 shared, 11 cheaper per output token (DeepSeek V4.1 Flash 1017: $0.29 vs $1.20)
- [Together AI](https://benchlm.ai/hosts/together): 10 shared, 0 cheaper per output token
- [Baseten](https://benchlm.ai/hosts/baseten): 9 shared, 2 cheaper per output token (GPT-OSS 120B: $0.50 vs $0.60)
- [Fireworks AI](https://benchlm.ai/hosts/fireworks): 7 shared, 0 cheaper per output token

## FAQ

### How much does the Nebius Token Factory API cost?

Nebius Token Factory bills per million tokens, with separate input and output rates for each model. On October 20, 2026 its output rates ran from $0.24 on Nemotron-3-Nano-30B-A3B to $15.00 across 27 paid models. Input is usually cheaper than output, so a long prompt with a short reply costs less than the output rate suggests.

### Which models can you run on Nebius Token Factory?

27 text models are listed, and 17 of them match a model we score. The highest-scoring is Kimi K3 at 79. The table on this page has every model with its input, output and cached-input rate, the context window, and the date we read the price.

### What are the alternatives to Nebius Token Factory?

OpenRouter lists 17 of the same models and charges less output per token on 11 of them. Novita AI and DeepInfra also overlap. Price is only part of the choice: speed, rate limits and which model snapshot a host serves differ too.

Canonical page: https://benchlm.ai/hosts/nebius