Jev pricing and API cost
Jev costs $0.042 per million input tokens, or $42 per billion, and output tokens are free, according to TypeSafe's Models page. You pay for what you send, meaning the state plus your questions, and nothing for the answers.
The published price
These figures are for jev-1.13.0, the model behind the jev-latest alias, as listed on the Models page when we checked it on September 25, 2026.
| Item | Jev 1.13 |
|---|---|
| Input price | $0.042 per million tokens ($42 per billion) |
| Output price | Free |
| Rate limits | 250,000 tokens per second, 1,200 requests per minute |
| Context length | 64k tokens per request; 32k for the state plus the longest question |
| Input types | Text only: string, JSON object, or array of text |
The same numbers appear in TypeSafe's launch post, which describes output tokens as "too cheap to meter." TypeSafe's homepage also lists $42 per billion input tokens.
What you are billed for
Billing is per input token. Every response includes a usage object with input_tokens and output_tokens, so you can log the cost of each call (API reference).
Input includes the state and every question's instructions and criteria. Jev reads the state once per request and answers all the questions against it (Models). That makes batching questions cheaper than sending them one at a time, because each separate request pays for the state again.
Worked examples
These use numbers published by TypeSafe. The arithmetic is ours.
- One support ticket. The quickstart example asks three questions about a short ticket and reports 392 input tokens (Quick start). At $0.042 per million, that is about $0.0000165 per call, or roughly $16.50 per million such calls.
- A long document, batched. TypeSafe's parallel-questions cookbook asks 13 questions about the roughly 54,000-character Wikipedia article on GDPR. One call with all 13 questions cost $0.000497; 13 separate calls cost $0.006090. TypeSafe reports that as 12.2x cheaper and 10.0x faster, with the same answers (Parallel questions). The cookbook ran on
jev-1.12at the same $0.042 input price. - A real-time loop. In the launch post's Doom demo, 10 queries per second cost about $7 per hour (launch post).
To estimate your own bill, multiply requests by average input tokens, then by $0.042 per million. For example, 1 million requests at 500 input tokens each is 500 million tokens, or $21.
How it compares with LLM pricing, per TypeSafe
The launch post puts frontier LLM input pricing at $0.20 to $10 per million tokens, with output tokens about 5x more expensive than input. TypeSafe's homepage claims Jev's input price is 238x lower than Claude Fable 5.1's, and that Jev is 444.6x cheaper on its published workflow evaluations.
Treat these as vendor claims. The launch post says the 444.6x figure is probably "on the higher end of real world gains," that the workflows were written by TypeSafe's own team, and that TypeSafe "can't prove it isn't subsidized," though it expects prices to go down, not up. Also, a lower token price only helps if Jev can answer your question well. See Jev vs LLMs for when it can.
Plans, limits, and what is not public
- Console pricing. The console at console.typesafe.ai was not publicly readable when we checked, so we could not confirm details such as free credits, minimum spend, or payment terms. The public docs do not mention a free tier.
- Access. The launch post says Jev is in early access and that TypeSafe is bringing developers off a waitlist.
- Rate limits can change. The Models page warns that limits are "adjusting dynamically" and can change without notice. Going over a limit returns
429 Too Many Requests. - Enterprise. Higher limits are available on custom and enterprise plans, and zero data retention is offered to enterprise customers (Legal). Contact sales@typesafe.ai for those prices, which are not published.
Keeping costs low
- Put all questions about one state into a single request (Speculative fan-out).
- Send only the fields a question needs. TypeSafe notes that unrelated detail also lowers accuracy (Jev 1.13 jaggedness).
- Do counting, math, and date comparisons in code rather than asking the model.
- Use Jev to route work so expensive LLM calls only happen when needed. See Routing and orchestration for examples.
Questions
How much does the Jev API cost?
Jev 1.13 costs $0.042 per million input tokens ($42 per billion). Output tokens are free.
Is there a free tier for Jev?
The public docs do not mention one, and the console's pricing page was not publicly readable when we checked. Sign in to the TypeSafe console or contact TypeSafe to confirm current terms.
Do more questions cost more?
Slightly. Each question's instructions and criteria count as input tokens, but the state is paid once per request, so adding questions to one call costs far less than sending separate calls.
Are output tokens charged?
No. TypeSafe charges only for input tokens. The usage field still reports output tokens so you can see them.
What happens if I exceed the rate limit?
The API returns 429 Too Many Requests. TypeSafe's SDKs retry with backoff by default; if you call the HTTP API directly, retry with exponential backoff.
Will Jev's price go up?
TypeSafe says it expects its pricing to go down, not up, but also that it cannot yet prove the current price is not subsidized.
More guides: What is Jev?Jev vs LLMs: when to use whichHow to use Jev: a quickstartJev and MCP: using Jev as a Model Context Protocol toolHow to use Jev with Claude CodeJev in an agentic harness: where it fits in an agent loopJev for orchestration: routing requests, models and skillsJev as a judge: evaluating LLM and agent outputsJev for SEO and GEOJev for ads and ad reviewJev for marketing and lead scoringIs there an open source Jev?Can you fine-tune Jev?Designing questions for JevWhat is RLCD?Jev statistics