Baseten Pricing
Per-token pricing for 23 open-source LLMs on Baseten — a model-deployment platform offering serverless inference for frontier open models.
Last updated: Nov 21, 2026
Get our weekly newsletter on pricing changes, new releases, and tools.
Provider | Model | Context | Quant | Input/1M | Output/1M | Cached input/1M |
|---|---|---|---|---|---|---|
16k | — | $0.010 | $0.000 | $0.00 | ||
16k | — | $0.035 | $0.000 | $0.00 | ||
groundhog-1 | 16k | — | $0.035 | $0.000 | $0.00 | |
bison-1 | 33k | — | $0.100 | $0.000 | $0.00 | |
16k | — | $0.100 | $0.000 | $0.00 | ||
128k | fp4 | $0.100 | $0.500 | $0.10 | ||
DS | deepseek-ai/DeepSeek-V4-Flash-0731 | 1049k | fp32 | $0.130 | $0.260 | $0.03 |
Z | 1049k | fp8 | $0.150 | $0.500 | $0.03 | |
Nemotron 3.5 Super VL | 1049k | fp4 | $0.200 | $0.500 | $0.04 | |
DS | 1049k | — | $0.300 | $1.200 | $0.01 | |
DS | deepseek/deepseek-v4.1-flash-1017 | 1049k | — | $0.300 | $1.200 | $0.01 |
1049k | fp4 | $0.400 | $1.200 | $0.04 | ||
DeepSeek V4.1 Flash 1017 Fast | 1049k | — | $0.600 | $2.400 | $0.01 | |
DS | 1049k | — | $0.600 | $2.400 | $0.01 | |
NV | 203k | fp4 | $0.600 | $2.400 | $0.12 | |
1049k | fp4 | $1.250 | $4.250 | $0.15 | ||
DS | deepseek/deepseek-v4-pro-0813 | 1049k | fp4 | $1.320 | $3.960 | $0.13 |
Z | 1049k | fp8 | $1.400 | $4.400 | $0.14 | |
Z | 1049k | fp4 | $1.400 | $4.400 | $0.14 | |
1049k | fp8 | $2.000 | $10.000 | $0.20 | ||
Z | 1049k | fp8 | $2.100 | $6.600 | $0.21 | |
Z | 1049k | fp8 | $2.100 | $6.600 | $0.21 | |
1049k | fp8 | $3.000 | $15.000 | $0.30 |
Pricing from the Baseten Model APIs, updated daily. Prices per 1M tokens.
About Baseten
Baseten is a model-deployment and inference platform — known for Truss and its dedicated, autoscaling deployments — that also offers serverless, pay-per-token Model APIs for popular open models like gpt-oss, GLM and Kimi. The API is OpenAI-compatible.
Teams typically use Baseten's serverless tier (priced below) for variable traffic and move to dedicated deployments for steady or latency-sensitive workloads. The table shows the models Baseten currently serves with public per-token pricing, the quantization each price applies to, and context length.
Baseten Pricing FAQ
How much does Baseten cost?
Baseten charges per token for its serverless Model APIs, with input prices for open models starting around $0.10 per million tokens. See the live table above for current per-model input and output prices.
Which models does Baseten support?
Popular open-source LLMs including gpt-oss, GLM and Kimi. The table above lists the models Baseten currently serves with public per-token pricing. Baseten also lets you deploy any custom model via Truss.
What is the difference between Baseten serverless and dedicated?
Serverless Model APIs are pay-per-token with no commitment, good for variable traffic. Dedicated deployments give you reserved, autoscaling GPU capacity for steady or latency-sensitive workloads.
Is Baseten OpenAI-compatible?
Yes. Baseten exposes an OpenAI-compatible API for its Model APIs, so switching from another provider is usually just a base-URL and API-key change.
