Jev pricing
TypeSafe’s public models page lists Jev 1.13 at:
- $42 per billion input tokens
- $0.042 per million input tokens
- Output tokens: free (“too cheap to meter” in the launch post)
Only input tokens are billed. The API still reports usage.output_tokens.
Vercel’s AI Gateway model card for Jev repeats $0.042 / 1M input tokens and no output charge. Gateway billing, budgets, and any platform fees are Vercel’s; read their invoice, not this paragraph.
What is not in the public docs
The models page does not, in the snapshot we used, describe a self-serve free tier, trial credits, or committed-use discounts. Higher rate limits are “available on custom and enterprise plans” via sales@typesafe.ai. We will not invent starter quotas.
The launch post says TypeSafe cannot prove the price is not subsidized and expects prices to go down, not up. That is their commentary.
Cost control that is documented
- Batch questions: state tokens are ingested once per request.
- Filter state: unused context costs money and, on 1.13, accuracy.
- Pin versions so a silent alias move does not invalidate a cost model you were not watching.
If a third-party blog quotes a different sticker price, ignore it until docs.typesafe.ai/models changes.
Sources
Public TypeSafe or adjacent documentation only. No private claims.