Tool-routing models

WebBrain Compass

Models of about 2–4B parameters trained to choose WebBrain’s next browser tool and its arguments in Compact mode: answer, ask for clarification, make a grounded tool call, or escalate. Plus a free hosted route, WebBrain Compass 1.0.

Models on Hugging Face

9 → 16strict matches /89, Tiny v2 vs MiniCPM5-2B

21 → 41loose matches /89, Tiny v2 vs its base

~1/10the size of Qwen3.8-27B, one strict match behind it

73.0%Sonnet match for hosted Compass 1.0, $0.12 per run

Why it matters

Most steps in a browser task are routine: read the page, click the right button. A small router that gets those right lets the big model do only the hard parts, and lets routing run on the edge or the device.

The models

Compass Tiny v1 (2.6B, from LiquidAI LFM2.5-2.6B) matched Qwen3.6-35B-A3B on expected tool names: 42/89, against 34 for its base and 12 for Qwen3.5-9B.

Compass Tiny v2 is a LoRA fine-tune of OpenBMB MiniCPM5-2B, trained for one epoch on 7,311 reviewed examples. It doubled its base on strict routing (9 → 16 of 89) and loose routing (21 → 41), one strict match behind Qwen3.8-27B (17/89) at roughly one-tenth the parameters.

Compass Tiny XS v3 moves to the 1.7B Spark-X2.5 base, with ONNX builds for the browser.