Ex-ChatGPT team launches Jev, a decision-only model claiming 20-200x speed at a fraction of the cost

TypeSafe AI has released its first model, Jev—a frontier model that "only makes decisions, never writes text": given an application state and a set of typed questions, it directly outputs options, scores, and probabilities for code to execute, generating no text at all. The team was co-founded by researchers behind ChatGPT (the community widely calls the founder a ChatGPT co-inventor/co-author; he spent two years in stealth developing a new training method called RLCD), and the model is positioned as "composable intelligence" for decision optimization rather than a better chatbot. Officially, it is claimed to be 20–200x faster and 40–400x cheaper than existing LLMs (output tokens are free), and described as one path toward AGI. Multiple posters highlighted its System 1 positioning: let fast, cheap models handle small decisions while reserving expensive System 2 models for complex reasoning.

Confirmed

• Jev is live on Vercel AI Gateway, open to all users, with zero data retention (ZDR) support; input is priced at $0.04/million tokens and output tokens are free (m8, m12, m19)
• Mechanically it returns structured decisions instead of text, supporting use cases like classification, routing, rubric scoring, tool selection, and guardrails (m3, m17)
• @Sam Witteveen tested it and confirmed it specializes in fast System 1 classification rather than long-chain reasoning (m2); @Cagnazzo82 demoed Jev playing Subway Surfers in real time to showcase its real-time decision-making (m1)
• Developer grichadev tested it in a safety pipeline: over 5x cheaper than previous approaches, faster, and notably more accurate (m13)

Unconfirmed

• The claimed "20–200x faster, 40–400x cheaper" figures come from the vendor; apart from grichadev's single-scenario test, broad third-party validation is still lacking (m4, m13)
• The founder's "ChatGPT co-inventor" claim is self-described, and the community holds differing versions of the accurate characterization (m4, m6, m15)
• Technical details of RLCD as a "new reinforcement learning paradigm" have not been sufficiently disclosed (m6, m18)

Why it matters

• If the cost and speed claims hold, many small decision tasks currently handled by LLMs generating token-by-token then parsed (classification, routing, guarding) could migrate to purpose-built models like this, significantly cutting AI application costs and latency—real pipelines like Sentry's have already started adopting it (m13)
• It represents the "models are more than chatbots" route: embedding structured, composable outputs directly into the software decision layer, and ecosystems like LangChain have already followed up with usage guides (m17)

2026-09-17 ~ 2026-09-19 · 63 related posts

Full story(5 episodes)→

• Episode 1: TypeSafe exits stealth with decision model Jev and RLCD training method(2026-09-16, 86 posts)
• Episode 2: TypeSafe AI launches Jev, a dedicated evaluation model showing major speed and cost gains in tests(2026-09-16, 8 posts)
• Episode 3: Vercel fx to adopt Jev safety reviewer, up to 18x faster(2026-09-17, 3 posts)
• Episode 4: Ex-ChatGPT team launches Jev, a decision-only model claiming 20-200x speed at a fraction of the cost(2026-09-17, 63 posts)
• Episode 5: TypeSafe Jev's LLM Probability Trick Is Easy for Giants to Copy(2026-09-18, 2 posts)

• TypesafeAI ships Jev, a zero-shot classifier that cuts chat-data labeling costs — yenkel · 2026-09-17
• Jev Debate: Engineers Forget Encoder-Only Classifiers Have Existed for Years — brandon_galang · 2026-09-17
• typesafeai's Jev classifier lands on AI Gateway with zero-data retention — cramforce · 2026-09-17
• Typesafe AI's Jev lands on Vercel AI Gateway for millisecond agent decisions — cramforce · 2026-09-17
• [source] New Model Jev Runs Security Pipelines 5x Cheaper and Faster, Devs Say — zeeg · 2026-09-17
• Hands-on with Jev: a classifier model to replace LLM-as-a-judge and route agents — doesdatmaksense · 2026-09-17
• Jev buzz reveals many don't know encoder-only classifiers have existed for years — RichmanRonald · 2026-09-17
• Dev says Jev classifies tweets 6X faster and ~40X cheaper than fastest LLM — altryne · 2026-09-17
• Switching tweet classification to Jev: 6x faster, ~40x cheaper than fastest LLMs — altryne · 2026-09-17
• [source] Jev evaluation model now available to everyone via Vercel AI Gateway at $0.04/1M tokens — jarrodwatts · 2026-09-17
• Six real-world uses for TypeSafe's Jev judgment model, 4x faster than Gemini in evals — HamelHusain · 2026-09-18
• Six Real Uses for Jev, the Judgment Model: From Fact-Checking Scripts to Debugging Agents — HamelHusain · 2026-09-18
• Self-Proclaimed ChatGPT Co-Inventor Launches Jev, Claims 200x Speed at 1/400 Cost — iamrobotbear · 2026-09-18
• Jev hype reveals two things: weak builder skills and the case for renting inference — edgarpavlovsky · 2026-09-18
• TypeSafe, founded by a ChatGPT researcher, launches Jev — a decision engine that can't write — Scobleizer · 2026-09-18
• TypeSafe's structured evaluation model Jev goes live on Cloudflare AI Gateway — michellechen · 2026-09-18
• 48 hours old, 724 ads broken down in 40 seconds: Jev sweeps the dev world — Scobleizer · 2026-09-18
• ChatGPT co-inventor launches Jev model with AI detector API 100x cheaper than Pangram — gaganghotra_ · 2026-09-18
• Jev Ditches Autoregression: A Model That Only Outputs Structured Decisions — karminski3 · 2026-09-18
• Jev: A Non-Autoregressive Decision Model That Outputs Structured JSON in 70ms — karminski3 · 2026-09-18
• Jev, the non-chatbot model, returns decisions in 70-500ms at $0.042/M input tokens, free output — adrianscottcom · 2026-09-18
• Jev and System One Models: fast decision-only AI that outputs typed answers, not text — blaizedsouza · 2026-09-18
• Jev, a 'System One' model by Typesafe, launches on OpenRouter with typed decisions instead of text — majidmanzarpour · 2026-09-18
• Dev: Jev could run cheaply in browsers — classification tasks were shoehorned into LLMs — GabGarrett · 2026-09-18
• TypeSafe launches Jev, a "System One" model that answers typed judgments in ~100ms — aronchick · 2026-09-18
• ChatGPT co-inventor launches Jev model claiming 20-200x speed and 40-400x cost cuts — multiply_matrix · 2026-09-18
• Jev, a new 'decision' model from a ChatGPT co-creator, plays Subway Surfers in real time — Cagnazzo82 · 2026-09-18
• ThursdAI Podcast Dives Into Jev Real-Time Models and Computer-Use Agents — altryne · 2026-09-18
• ThursdAI: TypeSafe's non-LLM Jev hailed as a ChatGPT moment, Dario proposes three ways to pace frontier AI — altryne · 2026-09-18
• A day with Jev: a fast classifier, best suited for Computer Use and game AI — jiayuan_jy · 2026-09-18
• After a day with Jev: a blazing-fast classifier, not a GPT replacement — jiayuan_jy · 2026-09-18
• Jev Launches as a Decision-Only Model: 20-200x Faster, 40-400x Cheaper Than LLMs — thisiskp_ · 2026-09-18
• TypeSafe Jev returns typed decisions, not text: 70ms latency, free output tokens — mostly_idempotent · 2026-09-18
• Stop using LLM prose for routing: Jev returns structured decisions at $0.042 per million tokens — mikegiannulis · 2026-09-18
• A 20-cent AI that can't write a sentence caught 13 requests our classifier missed — mikegiannulis · 2026-09-18
• Replaying 3,300 Real Production Decisions to Test Small Model Jev — mikegiannulis · 2026-09-18
• Offline replay of 3,300 production decisions validated Jev routing in an afternoon for 20 cents — mikegiannulis · 2026-09-18
• Before buying a smarter model, check you asked the wrong job: Jev test wrap-up and rollout rules — mikegiannulis · 2026-09-18
• Same Material, Better Order: Reranking Lifts Correct-Passage Hit Rate to 90% — mikegiannulis · 2026-09-18
• 13 'Please Text Me, No Calls' Messages Misrouted — Small Model Caught All of Them — mikegiannulis · 2026-09-18

10 near-duplicate retellings: threepointone · hardimanjames · HankYeomans · multiply_matrix · multiply_matrix · multiply_matrix · multiply_matrix · GabGarrett · multiply_matrix · hardimanjames