AI
Fewer Tokens, Better Action: GPT-6 Astra Robot Agents with 14% Higher Success Rate but 65% Fewer Tokens
Publication: Hugging Face Daily Papers
Author: Not available in collected metadata
Source published: 30/09/2026, 20:00:00 UTC
Briefing created: 03/10/2026, 03:40:05 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 02/10/2026, 09:18:24 UTC.
Briefings
A new framework called PyRUA-Lean improves robot agent success rates by 14% while reducing token usage by 65%. The study compares this approach against a standard tool-calling baseline using GPT-6 Astra agents across 700 simulated tasks from LIBERO-PRO, RoboTwin 2.0, and RoboCasa365. Key findings show that PyRUA-Lean achieves 71.7% success compared to 63.1% for the baseline, utilizing significantly fewer LLM calls and input tokens on successful instances.
Key points
- PyRUA-Lean achieves a 14% higher success rate (71.7% vs 63.1%) compared to a tool-calling baseline using GPT-6 Astra agents.
- The framework reduces input tokens by 65% on instances solved by both agents, using 49% fewer LLM calls.
Context and limitations
Results are based on 700 simulated task instances from LIBERO-PRO, RoboTwin 2.0, and RoboCasa365, with specific success rates and token savings reported only for instances solved by both agents.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 03/10/2026, 03:40:05 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeicknp64sis544bgpnjwo7p75cd7shoqdzdahdlmafk4gypgrpww3q