GLM 5.2 Orchestrator Enters Computer Harness
henrymodis · x · 2026-07-09
The post shares a research preview from Perplexity Computer: they applied post-training to a version of GLM 5.2, enabling it to handle orchestration within the Computer harness and escalate to a stronger frontier model when necessary.
The original post claims that when paired with an advisor, this orchestrator model achieves performance close to Opus 4.8 levels at a fraction of the cost.
Related event: Perplexity Releases GLM 5.2 Orchestrator Model Preview(7 posts)→
More from coding & agent
- Kolab Benchmark Uses Pro Screen Recordings to Show AI Agents Overestimate Collaboration Quality — minnesotanlp · 2026-10-27
- Dev argues Claude burns far more tokens than GPT models for the same coding work — haider1 · 2026-10-27
- Stop building agents that can't change models: stable contracts, external state, real evals — rseroter · 2026-10-27
- Cheap models can cost more in agent harnesses than frontier ones, researcher warns — vasuman · 2026-10-27
- Kanzo: A 5k-Line Rust Kanban Binary With MCP for Claude Code and Codex — allmudi · 2026-10-27
- 'A human approves everything' is a trap: approval fatigue and how to actually design safe agents — Efistoffeles · 2026-10-27