How much does the Jev API cost?
TypeSafe’s model documentation lists $42 per billion input tokens, equivalent to $0.042 per million. Output tokens are free. Price checked September 27, 2026; verify the provider’s current rate before budgeting.
Use billed input tokens for the complete request, including the state and questions. A request is not necessarily one decision: multiple questions may share a state.
Model: jev-1.13.0. The published documentation also lists rate limits that may change dynamically; price alone does not guarantee capacity.
Inspect the source ↗What does Jev cost per month?
| Requests / month | Input tokens / request | Monthly input tokens | Input charge |
|---|---|---|---|
| 10,000 | 1,000 | 10 million | $0.42 |
| 100,000 | 1,000 | 100 million | $4.20 |
| 1,000,000 | 1,000 | 1 billion | $42.00 |
| 1,000,000 | 4,000 | 4 billion | $168.00 |
Calculated as requests × billed input tokens per request ÷ 1,000,000 × $0.042. These examples exclude taxes, other services, retries, and any separate commercial terms. They are arithmetic scenarios, not observed customer bills.
Is self-hosting Laya free?
| Cost component | Jev API path | Laya self-hosted path |
|---|---|---|
| Inference | Input-token charge | Compute rental or owned hardware |
| Idle capacity | No idle-server cost in this token calculation | Depends on whether the machine stays running |
| Operations | Integration, retry and monitoring work | Runtime, model updates, capacity, monitoring |
| Other costs | Network and application services as applicable | Storage, network, power and engineering time |
Laya’s public code license does not create a per-token API fee for local inference. It also does not make hardware and operations free. Review checkpoint terms separately for your intended use.
When could self-hosting cost less?
Assume, purely for calculation, that self-hosted infrastructure costs $20/month and each Jev request uses 1,000 billed input tokens. Jev input cost is $0.000042/request. Dividing $20 by that rate gives about 476,190 requests/month.
At low usage, a cheap API can be less expensive than an always-on server. Self-hosting may still be justified by data control, customization, offline use, or an existing machine—but it should not be chosen on the assumption that “open source” always means lower total cost.
See measured hosting configurations →Sources & review notes
Source review: September 27, 2026. Product documentation can change. Atlas has not rerun model inference. Third-party measurements, where included, are attributed to their authors.
- TypeSafe model documentation — input-token price, free outputs, and current model
- Laya code license — Apache-2.0; hardware and operating costs are separate