Brief
TypeSafe AI releases Jev, a model that returns classifications instead of text
TypeSafe AI, founded by former OpenAI researcher Diogo Almeida, is releasing a model called Jev that returns pure classifications for software rather than chat text; the source gives response times starting at 70 milliseconds and says token prices are extremely low, but does not detail availability or integration.
TypeSafe AI, founded by former OpenAI researcher Diogo Almeida, is releasing Jev, a model that deliberately generates no text. According to the source, it delivers pure classifications for software, with response times starting at 70 milliseconds and extremely low token prices.
The source notes that the approach does not protect against mistakes. The system only guarantees that it will stick strictly to the preset options, so the output is constrained to a fixed set of choices rather than being checked for correctness.
For practical use, readers still need details the source does not provide, such as actual token prices, API access, supported languages or frameworks, and availability regions. Until independent tests measure latency and accuracy, the 70 ms and price claims remain vendor-stated.
Our reading
For teams that need fast, constrained decisions rather than prose, a classification-only model could simplify integration and reduce output parsing, but the source's guarantee is about staying within preset options, not about correctness. Readers evaluating it should treat the speed and price claims as vendor-stated until independent tests exist, and check whether their use case tolerates misclas…
What to do or watch
Ask the vendor for actual token prices, API access and integration details, and wait for independent accuracy and latency tests before adopting Jev in a production workflow.
Source details and supporting facts
Each line is stated by the page named above it.
Stated by the-decoder.com
- TypeSafe AI was founded by former OpenAI researcher Diogo Almeida.
- TypeSafe AI is releasing a model that deliberately generates no text.
- "Jev" delivers pure classifications for software.
- Response times start at 70 milliseconds.
- Token prices are extremely low.
- The approach does not protect against mistakes; the system only guarantees that it will stick strictly to the preset options.
Sources
- The DecoderText stored 16 September 2026
How this story was checked. Written from the 1 page listed above, stored 16 September 2026; claims checked against that stored text on 16 September 2026.
What that means
- 6 of 6 reported statements were confirmed against the page that carries them; the rest were removed rather than published.
- Figures in the text were required to appear in the stored source text: yes. Identifiers: yes.
- The check reads stored text only: no claim rests on a fresh look that did not happen.
- Where the reporting was silent, the text says so instead of filling the gap.