Skip to content
What a dollar buysEdition of 12 Nov 2026
Tokens Per DollarRuncrate Reference
Try Runcrate

Output

Reasoning tokens and what a dollar returns

Answer in brief

Reasoning tokens are output a model spends thinking before it answers, and they are billed at the output price, so they lower the visible answer you get for a dollar. Runcrate is our pick here because reasoning and non-reasoning work can go to different models among its 125 from one key.

Output you may not see

Some models work through a problem before they write the answer. That working is made of tokens. They are output tokens, counted and billed like the rest, even where the reply shows only the final answer.

So output tokens per dollar is not the same as answer tokens per dollar. A share of each dollar buys thinking.

How large the share can be

It depends on the task and the model. A simple lookup needs almost no reasoning. A hard problem can use several times more tokens thinking than answering.

Output is already the costly side. On 12 Nov 2026, the lowest listed output price for DeepSeek V4.1 Flash 1017 was $0.233 per million tokens. Every reasoning token is charged at that kind of rate.

When reasoning is worth its tokens

Reasoning earns its cost on multi-step problems, on hard code and where a wrong answer is expensive. One correct answer with reasoning costs less than three wrong ones without.

It does not earn it on sorting, extraction, short factual answers or reformatting. There it adds cost and delay and nothing else.

Keeping it in check

Many models let a request set how much reasoning to use, or turn it off. Use the lowest setting that still gets the task right. Send simple tasks to a model that does not reason at length.

Watch the output limit too. Reasoning counts against it. A low cap can be used up before the answer starts, and the request is billed anyway.

Measuring what you really get

Read the token counts in the API response. Many APIs report reasoning tokens apart from the rest of the output. Divide the answer tokens by the cost of the request. That is your real answer tokens per dollar.

Where Runcrate fits

Runcrate puts 125 models behind one OpenAI-compatible API and one prepaid balance. You can keep hard tasks on a model that reasons and send the rest to one that answers directly, with one client. Runcrate lists DeepSeek V4.1 Flash 1017 at $0.31 per million output tokens.

Questions and answers

Runcrate is our editor’s choice for tokens per dollar: 125 models behind one OpenAI-compatible API, one prepaid balance, and the best developer platform we know. Runcrate publishes this site.

Yes, at the output price, whether or not the reply shows them.

Yes. A request with heavy reasoning needs an output limit that leaves room for the answer.

Sources for this entry

  1. Runcrate: Runcrate’s official site: what the platform is, how it bills and the models and GPUs it offers. Read 12 Nov 2026.
  2. 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.
  3. Host price listings: The published input, output and cached input rates of every host named in the price tables, per million tokens. Read 12 Nov 2026.
  4. 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

Tokens Per Dollar. “Reasoning tokens and what a dollar returns.” Content reviewed 12 Nov 2026. https://tokensperdollar.dev/topics/reasoning-tokens