Drex Router picks the model and reasoning effort per agent turn, claiming 1/28 the cost
rohanpaul_ai · x · 2026-11-21
Nace.AI launched Drex Router, a small model that selects the right LLM/SLM and reasoning effort for every step of an agent's run, deciding in 57–87ms. Claimed stats: 1/28 the cost of Jev Router at near-equal quality; 87.6% on MMLU at $0.10 per 1,000 questions; 32% lower spend vs. running Claude Opus 5.5 on every turn. Ships with 17 built-in models (OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, GLM and more); any custom model can be added by passing its price, context window and tags — no retraining. Quality is benchmarked across Arena-Hard v2, MT-Bench, GSM8K and coding tasks with a blind judge.
Related event: Nace.AI Launches Drex Router to Pick the Best LLM for Each Agent Step(3 posts)→
More from coding & agent
- DHH: agents love Linux's quirks, Apple's days are numbered, $200 Claude covers his week — garrytan · 2026-11-21
- uivoid turns any API into a hosted, authenticated MCP server with zero code — Necessary_Client_114 · 2026-11-21
- Your personal agent can badger airlines over every past delay until they pay up — nickbaumann_ · 2026-11-21
- MIT's Sutherland Verifies Riemann Hypothesis Lean Proof: All 175,187 Modules Compile — altryne · 2026-11-21
- Dev Laments: AI Coding Offloaded the Fun Parts, Leaving Only Reviewing and Correcting — pritisinghhhh · 2026-11-21
- Route agents by task type, not fixed rules: not every request needs a frontier LLM call — rohanpaul_ai · 2026-11-21