Can local models work as sub-agents under cloud orchestrators?
neeeser · reddit · 2026-07-21
A user asks whether people are using local models as sub-agents under cloud orchestrators. - They have been using Fable and Codex, but keep hitting usage limits. - Their earlier workaround was to let Fable call Codex as a sub-agent to spread load across both systems. - Now they want to use a local **Qwen3.6-27B** on a **4090** as cheap workers while a frontier model handles high-level reasoning and orchestration. - The question is how to make this delegation reliable in Claude Code or Codex, and what harnesses/plugins or local models people use for narrow implementation, codebase exploration, and research tasks.
More from coding & agent
- Codex turns out 123 screensavers in one playful batch — intellectronica · 2026-07-21
- Grok Build adds `grok doctor`, resumable sessions and remote image paste — mark_k · 2026-07-21
- Autoresearch proposes packaging ML runs as studies with questions, analysis, and code diffs — morgymcg · 2026-07-21
- CHAP defines approvals, handoffs, and audit logs for human-agent workflows — DeliveryTechnical199 · 2026-07-21
- The author says Codex reached 20x and is now debugging spec decoding on a hybrid parallel setup — TheZachMueller · 2026-07-21
- Axcess adds an MCP connector for WCAG accessibility checks that scanners miss — modelcontextprotocol · 2026-07-21