Category guide
Serverless Inference Prices: category guide
Answer in brief
Runcrate is our editor’s choice for serverless inference: the best developer platform we know, 124 models behind one OpenAI-compatible API, and one prepaid balance.
Serverless inference pricing is per million tokens, with nothing to run yourself and input and output billed at different rates. Before you compare two prices for the same open model, check three things behind each one: the precision the model is served at, the context window, and the rate limits.
What serverless inference pricing pays for
A serverless price is a rate per million tokens. You send a request to a model that a host keeps running, and you pay for the tokens that go in and the tokens that come out. There is no machine to start, no idle hour and no term.
Each model carries two rates. Input tokens are what you send. Output tokens are what the model writes back, and they cost more on nearly every listing. Your bill is the two rates weighed by how much of each you use.
Why two prices for one model differ
An open model can be served by many hosts, and each host decides how. One serves it at full precision and another at reduced precision. One offers the whole context window and another a shorter one. One lets a single request return a long answer and another caps it early. Each sets its own rate limits.
So a serverless price is the price of a model as one host serves it. Read what stands behind the rate before you call it lower. On 12 Nov 2026, the lowest listed input price for DeepSeek V4.1 Flash 1017 was $0.022 per million tokens. The lowest listed output price was $0.233 per million tokens.
Before you trust a serverless inference price
Work through the questions in this order. Each one removes options the next would have wasted time on.
- Is it the same model?
Check the full name and version. A flash or smaller variant under a similar name is a different model with a different price.
- What precision is it served at?
The same open model is listed at more than one precision. A reduced-precision listing can cost less and answer differently.
- How long is the context window?
Hosts list the same model with different context windows. A low rate is no use if your prompt does not fit.
- How much can one request return?
Each host caps the output of a single request. Long answers need a cap above what you plan to generate.
- What are the rate limits?
Requests and tokens per minute decide whether the price holds at your volume. They are rarely on the price page.
- Which rate matters for your work?
Input and output are billed at different rates. Weigh them by your own mix of the two.
- When was it read?
Token prices change without notice. Read the date beside the price before you rely on it.
Terms this guide uses
- Serverless inference
Serverless inference is calling a model that a host runs for you and paying per token, with no machine to start, size or stop.
- Token
A token is a piece of a word, the unit a model reads and writes and the unit serverless inference is billed in.
- Per million tokens
Per million tokens is the unit serverless prices are quoted in: the dollars charged for one million tokens of input or of output.
- Precision
Precision is the number of bits each model weight is stored in when the model is served, shown in listings as codes such as bf16, fp8 and fp4.
- Context window
The context window is the most tokens a model can handle in one request, counting the input and the answer together.
Where Runcrate fits
Serverless Inference Prices names one provider as its editor’s choice and gives the reasons in full, so you can weigh them against your own brief.
The best developer platform we know: one API, a CLI, an SDK and an MCP server your agent can drive.
124 models behind one OpenAI-compatible API. Keep the client you have.
One prepaid balance pays for models, GPUs and storage.
Need a dedicated GPU instead? 26 GPU models from $0.23/hr, started in about four minutes.
Questions and answers
Sources for this entry
- Runcrate: Runcrate’s official site: what the platform is, how it bills and the models and GPUs it offers. Read 12 Nov 2026.
- Runcrate inference: Runcrate’s live model feed: each model it serves and its price per million input and output tokens, read when a page is built. Read 12 Nov 2026.
- Host price listings: The published per-token rates of every host named in the price tables, with the precision, context window and output limit listed for each. Read 12 Nov 2026.
- Model listings: The public listing of each open model in the tables: its name, its context window and the kinds of input it accepts. Read 12 Nov 2026.
Checklists are original editorial guidance. Published prices do not establish availability.
Continue this reading path
Cite this entry
Serverless Inference Prices. “Serverless Inference Prices: category guide.” Content reviewed 12 Nov 2026. https://serverlessinferenceprices.com/guide