Tiny 4B decision model drives agent actions without 500-token reasoning chains
MaziyarPanahi · x · 2026-09-26
The author demos NeoHorse-Jev-4B, a small System 1 model acting as the "what next?" step in agents: given a tiny delivery car, at each junction it receives road state plus valid turns and directly outputs one choice (north → east → north, parcel delivered) — no 500-token driving monologue.
He argues the pattern shines for tool routing, go/no-go checks, next browser actions, and ranking short lists: when your app already knows the valid options, feed a decision model the state + choices, let it score them, and execute the winner in code. Decide → act → repeat.
Related event: 4B Model Drives Mini Delivery Cart with Instant Intersection Decisions(2 posts)→
More from coding & agent
- Explosion launches Ellf directory: 52 NLP templates plus multi-agent annotation workflow — _inesmontani · 2026-11-11
- Meme: 'Stochastic AI Agent Architecture' Is Just a Markov Chain in Disguise — PrasoonPratham · 2026-11-11
- OpenCode adds Gemini 4 Carbon and Claude Fable 5.2 support — dean_rie · 2026-11-11
- OpenClaw v2026.11.3 patch fixes cert verification and credential exposure across 56 PRs — dean_rie · 2026-11-11
- Fable 5.2 captured 1 in 4 Claude replies on day one, then growth stalled — dean_rie · 2026-11-11
- QA engineer proposes MCP agent to automate ad postback checks before every app release — Soobhee_Chukwutosin · 2026-11-11