Real dev work fully local: Ornith 1.5 35B on 8GB VRAM hits 30-40 tok/s with OpenCode
Sensitive_Song4219 · reddit · 2026-09-23
After weeks of a 'local-AI-only' regime, the author reports solid agent coding on an RTX 5060 Mobile (8GB VRAM): Ornith 1.5 35b-a3b runs at 30-40 tok/s generation. Key points:
- Model choice: Qwen 3.8 27b remains the local coding star, but Ornith 1.5 35b-a3b is a strong 30b-a3b alternative — extra training gives it a 'High reasoning'-like depth, whereas Qwen 3.6 35b-a3b MoE only has on/off thinking and tends to underthink
- Harness: Ornith struggled with file writes in Pi; OpenCode via llama.cpp fixed it at 11k base context, and OpenCode enables seamless mid-chat switching to cloud models
- Tests: a web-OS demo worked with two bug-fix turns; fully local router-log troubleshooting identified a WPS-related WLAN crash in 10 minutes and autonomously built a log-scraper/monitoring app
Verdict: agentic coding on modest hardware is already quite usable.
More from coding & agent
- Pigment Agents now reads and posts in Slack, inheriting user permissions — mattturck · 2026-11-12
- Map of the agent economy: 50+ startups across payments, spend control, identity, memory — venturetwins · 2026-11-12
- OpenCode launches Browse, an on-device browser agent with local logins — Saul_Loveman · 2026-11-12
- VentureTwins maps the B2A economy: dozens of agent infra and app companies in first series installment — venturetwins · 2026-11-12
- Google releases 76-page 'Agents Companion' whitepaper on agent engineering — mdancho84 · 2026-11-12
- Spinning up containers for 20-line agent tools is backwards: Safescript compiles provable, 50ms-deploy apps — uriwa · 2026-11-12