Claude Code fan wants to go local: OpenCode vs Codex vs Claude Code pointed at a local endpoint
Generic_Name_Here · reddit · 2026-11-04
An AI image/video professional with multiple RTX Pro 6000s wants to move privacy-sensitive and token-heavy tasks to local models (Qwen 3.8 27B in LM Studio), but LM Studio lacks a Claude Code-style harness. Target tasks: whole-codebase security/efficiency analysis, captioning a 10k-image dataset, driving a ComfyUI instance, PDF editing, and web research.
He narrows candidates to OpenCode, Codex and Claude Code pointed at a local endpoint, also asking about Openhands and Pi — and whether the official OpenAI/Anthropic harnesses still require login/online connection when running against a local model.
More from coding & agent
- How LangChain optimized browser use for Deep Agents with Stagehand — LangChain · 2026-11-04
- Plane AI scores 165/165 on real workspace eval: 18s median latency, $0.10 per answer — JosephJacks_ · 2026-11-04
- SurrealDB 4 launches with SurrealDB Earth: serverless agent database on Cloudflare — ritakozlov · 2026-11-04
- Karpathy's "vibe coding" post hailed as a shoo-in for the test-of-time award — MillionInt · 2026-11-04
- OpenAI SDK ships official WebSocket helpers for Responses, GPT-Live and Realtime APIs — craigsdennis · 2026-11-04
- Pydantic AI adds ConversationStore to prevent concurrent chat saves from overwriting each other — samuelcolvin · 2026-11-04