Jev pricing is based on input tokens. The official model specification checked on September 20, 2026 lists Jev 1.13.0 at $0.042 per million input tokens; output tokens are free. This page explains that published model rate, not a promise of your account’s final invoice or access conditions.
What does the published Jev price cover?
The rate is $42 per billion input tokens, equivalent to $0.042 per million. Do not confuse the billion-token and million-token units. Input includes the context and questions sent to the model; use the response’s usage.input_tokens for measurement instead of estimating from characters. Output tokens can appear in usage even though their published model price is zero.
How do you calculate API cost?
Estimated model cost = requests × average input tokens per request ÷ 1,000,000 × $0.042. For example, 100,000 requests at 1,000 input tokens each total 100 million input tokens: 100 × $0.042 = $4.20. These are illustrative workloads, not measured token counts from our API calls.
Worked cost examples
All rows assume 1,000 billed input tokens per request, including questions. Values use USD; they are not automatic currency conversions or region-specific offers.
per 1 million input tokens · output tokens are free
| Requests | Input tokens / request | Total input tokens | Estimated model cost (USD) |
|---|---|---|---|
| 1,000 | 1,000 | 1,000,000 | $0.042 |
| 10,000 | 1,000 | 10,000,000 | $0.420 |
| 100,000 | 1,000 | 100,000,000 | $4.200 |
| 1,000,000 | 1,000 | 1,000,000,000 | $42.000 |
What can increase the total budget?
Longer context and more questions can increase input usage. Additional calls, evaluation runs, retries, and multi-stage workflows can increase the amount of processing; check the provider’s actual usage and billing records for failed requests. Hosting, retrieval, a separate generative model, network costs, and any applicable tax are outside these calculations.
Do rate and context limits affect price?
Limits constrain throughput and request size; they do not change the arithmetic above. The checked documentation lists 1,200 requests per minute and 250,000 tokens per second, with a warning that limits are changing. It also lists a 64k total input budget and a 32k budget for state plus the longest question. Read current account limits before planning capacity.
Is there a free tier?
This guide does not promise a free quota or unlimited access. Check the official console for your account’s current access, billing, and any credit terms. A public per-token rate alone does not establish a free tier or a service-level commitment.
How should you budget a real application?
Start with a representative sample, record usage.input_tokens, and calculate both typical and long-input cases. Budget for the entire workflow, not just one model request. Pin the model version when comparing results and recheck the published rate before making a cost commitment.
Continue reading
Overview · Use cases · Get started · Limitations