AutoEnvScaling: proposer agent builds its own RL environments, Qwen3.6-35B doubles Terminal Bench 4 score
jzl86 · x · 2026-10-15
Natasha Jaques' team introduces AutoEnvScaling, pushing multi-agent recursive self-improvement: a proposer agent builds complete Harbor sandbox environments to challenge a solver agent, learning from its successes and failures to update its own memories, skills, and tools — no human in the loop. Qwen3.6-35B-A3B doubles its score on Terminal Bench 4 and exceeds Claude Opus-5 on HiL-Bench, working both with a proprietary proposer and in full self-play with open-weight models.
Related event: AutoEnvScaling Lets Agents Build Their Own Training Environments(3 posts)→
More from coding & agent
- Creator builds custom GoPro ingestion tool with Codex: 4-camera array with WiFi trickle ingest — OpenAIDevs · 2026-10-15
- HybridAgent reasons in latent space, acts via text — 2x faster, 3x cheaper, open-sourced — zainhas · 2026-10-15
- Claude Code 2.1.300 ships 88 CLI changes, drops AskUserQuestion directive from system prompt — ClaudeCodeLog · 2026-10-15
- Tahoe AI opens Tara, a one-prompt plug-in connecting agents to its Tahoe-100M biology datasets — anshulkundaje · 2026-10-15
- sgrep cuts coding agent LLM costs 21.3% on SWE-bench versus ripgrep, fully local in Rust — udmrzn · 2026-10-15
- 10 GitHub repos that turn Claude Code and Codex into elite UI designers — udmrzn · 2026-10-15