AI Briefing
KO

Open-Source Router 'Brick' Released, Cutting LLM Costs by Up to 22x

·2026.07.06 19:08

Key point

Brick, an open-source router that analyzes query difficulty and capability to route to the optimal low-cost model, has been launched.

Details

Developed by the Regolo team, Brick is a Mixture-of-Models router that analyzes a prompt's capability requirements (coding, math, reasoning, creativity, planning, knowledge) and complexity, then forwards the query to the cheapest model capable of handling the task.

Key mechanism:

  • Capability classification: Classifies prompts across 6 dimensions (coding, math_reasoning, creative_synthesis, instruction_following, planning_agentic, world_knowledge).
  • Complexity assessment: A second classifier scores difficulty as Easy, Medium, or Hard.
  • Routing decision: Calculates a skill-distance score for each model to select the most cost-efficient model while maintaining quality.

Key features and benefits:

  • Cost savings: A real-world case study found that processing the same 1,000 queries could cut daily costs by approximately 79%, from $165 to $35.
  • Easy integration: Compatible with Claude Code or any client using an OpenAI-compatible API, and can run on CPU without requiring a separate GPU.
  • Open source: Released under the Apache 2.0 license, enabling free use.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.