Accessing the unofficial Jev documentation guide
Documentation access here means two different things: (1) TypeSafe’s official docs at docs.typesafe.ai, and (2) this independent unofficial guide at jev.pro. If they disagree, believe TypeSafe. Jev is TypeSafe’s System One decision model: unstructured state in, typed probabilistic answers out.
This page is for builders who already heard of Jev and need a reading order, not a credential. jev.pro is an independent unofficial field guide — not affiliated with TypeSafe AI. We do not sell, issue, or proxy API keys. Keys come from the TypeSafe dashboard after access at console.typesafe.ai. Public pages on this site are English only; we do not publish localized copies that could drift.
Steps
- Open the official introduction, quick start, API reference, primitives, and models pages on docs.typesafe.ai. That is the contract:
POST /v1/systemone,GET /v1/models, and the published model ids. - Read the unofficial mental model next: What is Jev and the documentation map. Use those pages to learn vocabulary (Choice, Score, Noul, floors, traces), not to invent extra HTTP verbs.
- When you are ready to call the API, follow how to get access, then store a key per API keys. Official HTTP auth is
Authorization: Bearer. Official SDKs readTYPESAFE_API_KEY. - Integrate with your first API call. Copy the request shape you settled in the official Playground; do not treat playground clicks as a published SLA.
- Operate last: pin a versioned id such as
jev-1.13.0before you tune floors, and keep a short learning path for the rest of the team. - If a rival handbook lists a different URL tree, treat it as demand radar. Cover ≠ clone. This site will not mirror competitor slugs.
Pitfalls
- Treating jev.pro titles that look like extra endpoints as official URLs. Many API pages here are application patterns around
POST /v1/systemone. There is no documented first-party/v1/roles,/v1/promote, or/v1/rate-limitprobe. - Using this guide as a key desk. We cannot mint, reset, or look up waitlist status. A 401 means a bad or missing key — fix the secret; do not retry 401.
- Pasting vendor speed or price as if we measured it. TypeSafe’s public models page claims $0.042 per million input tokens, output free, 250,000 tokens/second, and 1,200 RPM. Those figures can change; confirm on official docs.
- Mixing a TypeSafe bearer key with a Vercel AI Gateway base URL. Gateway auth is a Gateway credential or OIDC, and the model id is
typesafe-ai/jev— not a TypeSafe first-party alias. - Scraping jev.pro and citing it as TypeSafe. We are unofficial by design.
How the two surfaces stay useful
Official docs win on request and response fields, aliases, and legal text. This field guide wins on practitioner sequencing: which page to read first, which mistakes show up in production, and how to keep IAM, secrets, and promotion in your repo. The access hub is the cluster index. If you only need a credential path, skip this page and go to get access. If you only need header shape, go to authentication concepts.
Inventory stubs still exist elsewhere on the site. Prefer pages that cite official sources and label vendor claims. Do not wait on an unofficial CLI or a localized docs set from jev.pro.
See also
FAQ
Do I need an account to read jev.pro?
No. You need TypeSafe console access or a documented gateway credential to call Jev. Reading this guide does not grant a key.
Is the Playground documentation?
It is the official console UI for pasting state and mixing primitives. The contract is still the API reference.
Can I scrape this site as official?
No. We are unofficial by design. Quote TypeSafe when the fact is theirs.
What this page does not claim
This page does not claim that jev.pro is TypeSafe documentation, that we host a second API, or that we can issue keys. It does not freeze list prices or RPM. It does not promise a complete inventory of every official page.
Disclaimer
This is an independent unofficial site and is not affiliated with TypeSafe AI; official documentation is available at https://docs.typesafe.ai.