Jev doesn't generate tokens at all: why parallel decision evaluation differs from autoregressive LLMs
markjeffrey · x · 2026-09-20
- A widely shared thread explains the core difference between "LLMs vs. Jev": it's not that Jev is faster at generating — it doesn't generate text at all.
- A traditional LLM receives context and produces an answer one token at a time; even a small JSON output pays the autoregressive cost, since every token depends on prior ones.
- Jev receives the same context but evaluates predefined decisions directly, and independent decisions can all be evaluated in parallel.
- Example: an agent triaging a failed deployment must decide urgency, owning team, command riskiness, and task completion — independent judgments that map naturally to parallel evaluation instead of long generated reasoning.
Related event: Jev Gains Chinese Attention: Typed Probabilistic Decisions, No Text(2 posts)→
More from coding & agent
- Agent workflow: RFB scripts + tart macOS VM for autonomous UI iteration — craigbalding · 2026-09-20
- cairn-memory open-sourced: agent memory with inspectable source receipts to stop silent rewriting — sujingshen · 2026-09-20
- Jev Bets Agents Don't Need Big LLMs: Claims 20-200x Speed, 40-400x Cost Cuts — curiousperhaps · 2026-09-20
- Ex-Parse cofounder picks Opencode+Muse as favorite cheap non-frontier coding setup — alexandr_wang · 2026-09-20
- Giving Research Agents an Editable Sheet to Surface Conflicting Answers — truecakesnake · 2026-09-20
- Opencode Free Tier Acting Up? Retry Timer Balloons to 63,000 Seconds — Silver-Champion-4846 · 2026-09-20