Llama · Meta · AI Agent · Claude · Crypto Briefing
Meta and Carnegie Mellon study finds an RL-trained 8B model beats Llama-4-Maverick at tool selection
Compiled by KHAO Editorial — aggregated from 1 source. See llms.txt for citation guidance.
◌ Single Source
Meta Platforms logo (public domain) via Wikimedia Commons.
Key facts
- Qwen3-8B (RL-trained): 41.5% Qwen3-4B (RL-trained): 40.3% Qwen3-1.7B (RL-trained): 37.6% Llama-4-Maverick (~402B): 30.6%
- A Qwen3-8B model post-trained with reinforcement learning hit 41.5% on ToolBench G3, ahead of the roughly 402B-parameter Llama-4-Maverick at 30.6%
- The 1.7B Qwen3 variant reached a 37.6% success rate, and the 4B variant hit 40.3%
- The work appears in an arXiv paper titled “Training Small Language Models for Transferable Tool Selection,” listed as arXiv:2610.31483
Summary
A Qwen3-8B model post-trained with reinforcement learning hit 41.5% on ToolBench G3, ahead of the roughly 402B-parameter Llama-4-Maverick at 30.6%. A language model with 8 billion parameters outscored one roughly fifty times its size. Researchers from Meta and Carnegie Mellon University report that a small Qwen3-8B model, post-trained with reinforcement learning, selected the correct tools 41.5% of the time on the ToolBench G3 benchmark. The work appears in an arXiv paper titled “Training Small Language Models for Transferable Tool Selection,” listed as arXiv:2610.31483. The problem it tackles is a practical one for AI agents.