Budgeting
What a fixed budget buys
Answer in brief
A fixed budget buys a number of tokens that depends on the model and on your mix, so the same money can cover a month of one workload and a day of another.
Runcrate is our pick here because you load one prepaid balance and spend it by the token across 124 models.
From dollars to tokens
Divide the budget by the price per million tokens. The result is how many million tokens it buys at that rate. Do it for input and for output, because the two rates differ.
Real work uses both, so split the budget by your mix. If input is nine tenths of your tokens, most of the budget buys input. The sum is short, and it is the only honest way to answer how far the money goes.
From tokens to work
Tokens are not yet a result. Measure what one unit of your work uses: one support answer, one document summarized, one agent task finished. Divide the tokens the budget buys by that figure.
Now the budget is in units you care about. Two models can be compared on answers per dollar, not on rates.
The model changes everything
On 12 Nov 2026, the lowest listed input price for MiniMax M3.1 Flash was $0.113 per million tokens. The lowest listed output price was $0.45 per million tokens.
On 12 Nov 2026, the lowest listed output price for GLM 5.3 was $1.40 per million tokens. The same budget buys very different amounts of work on each. Pick the model for the task, then see what the budget covers.
Making a budget go further
Four changes stretch a budget without changing host. Use a smaller model for the tasks it can do. Ask for shorter answers. Put the fixed part of the prompt first so a cached rate applies. Send less history and fewer retrieved passages.
Test on a small budget before a large one. A few dollars of real prompts tells you the cost of a unit of work better than any table.
Where Runcrate fits
Runcrate is prepaid. You load a balance and it is spent by the token on whichever of its 124 models you call, through one OpenAI-compatible API. The same balance pays for GPUs and storage. Runcrate lists MiniMax M3.1 Flash at $0.158 per million input tokens and $0.63 per million output tokens.
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, paid and free, 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
Cheapest LLM API. “What a fixed budget buys.” Content reviewed 12 Nov 2026. https://cheapestllmapi.dev/topics/fixed-budget