Typesafe AI Launches Jev: A Decision Engine at 70-500ms and $0.042 Per Million Tokens
brandon_galang · x · 2026-09-17
Typesafe AI launched Jev, a generalized decision engine (not just a classifier) the author says occupies a unique spot on the cost-latency-intelligence-generality curve: 70-500ms per decision, $0.042/M input tokens, output free, $0.0001 per pass.
- The post maps the existing "Decision Tax": frontier LLMs are slow/expensive, small models are dumb, code is brittle, custom classifiers take weeks and rot
- Only frontier models out-judge Jev, at roughly 75x the cost and 25x the latency
- Because passes are cheap and fast, Jev works as a composable primitive: split one fuzzy question into ten small parallel judgments, each independently scored; run multiple passes over the same state with negligible latency
- Enables decision support across every branch, validation, and edge case with calibrated probabilistic judgments your code can branch on
Related event: ChatGPT co-inventor launches Jev, a decision-only model 200x faster(80 posts)→
More from coding & agent
- Nautilo pitches time-limited delegated E2E encryption and hierarchical memory for AI agents — Dan_Jeffries1 · 2026-09-17
- Nautilo uses memory namespaces so AI agents can't leak what they can't access — Dan_Jeffries1 · 2026-09-17
- Why Nautilo rebuilt chat: WhatsApp and Signal were never designed for agents — Dan_Jeffries1 · 2026-09-17
- Nautilo launches: a 100% open-source org harness with a personal 'Genie' agent for every user — Dan_Jeffries1 · 2026-09-17
- Open tutorial: training a search agent with GRPO, every rollout browsable and code open-sourced — dejavucoder · 2026-09-17
- Unsloth Desktop Launches: Train & Run 500+ Local Models, 70% Less VRAM — danielhanchen · 2026-09-17