A new kind of AI model from a ChatGPT inventor is thrilling developers
Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.
TypeSafe AI’s new non-language transformer model, Jev, outputs calibrated decision probabilities instead of text, making output tokens free and metering input tokens by the billion to run up to 18 times faster and 20 times cheaper than LLMs. For production systems, this completely eliminates hallucinations and provides genuine confidence scores for deterministic tasks like classification, model routing, and agent monitoring. You can now replace expensive LLM-as-a-judge guardrails with ultra-low-latency, highly accurate decision models at a fraction of the cost.
Jev outputs predefined calibrated probabilities instead of text, with free output tokens and input metered by the billion; Vercel reported 5–18x faster safety classification than an OpenAI model, and another test found Gemini only slightly more accurate while costing 10–20x more. For production agents, this makes bounded decisions like command safety, jailbreak monitoring, email classification, and model routing cheap enough to run inline, but it only works where you can define the decision space and act on confidence thresholds.