Jev vs GLM 5.3 at Chess: Checkmated in 29 Moves but 80x Cheaper Per Move
nutlope · x · 2026-09-19
The author pitted the specialized small model Jev against GLM 5.3 in chess: GLM 5.3 won by checkmate on move 29. Jev took 0.3s and under $0.0001 per move, while GLM 5.3 took 5.8s and $0.008 per move; the whole game cost 24 cents.
The takeaway is to play to each model's strengths:
- Fast, well-defined classification → specialized models like Jev
- Classification needing reasoning or lookahead → LLMs like GLM 5.3
- Real classification pipelines → hybrid, with Jev handling easy calls and an open model handling hard ones
The author argues the future is multi-model.
More from coding & agent
- Dev: Jev is best used to jumpstart training your own focused classifier — triviatroy · 2026-09-19
- OpenWA: Open-Source Self-Hosted WhatsApp API Gateway Hits 14k Stars — tom_doerr · 2026-09-19
- trycua open-sources CUA-S1-FORMS, first in a family of small computer-use 'System One' models — ycombinator · 2026-09-19
- A single prompt builds a playable 3D platformer as Astra orchestrates Thrixel's asset pipelines — RanaHanocka · 2026-09-19
- OpenDDE plugins to bring structure prediction and protein design into Codex and Claude Code by early October — AllThingsApx · 2026-09-19
- World models vs RAG: why agents need state tracking, not just retrieval — techNmak · 2026-09-19