Testing local Ornith 1.5 35B: solid performance but hardware is the bottleneck
DanWahlin · x · 2026-08-27
The author ran Ornith 1.5 35B (Q5 quantized) locally via the GitHub Copilot app on an M3 Max (64GB) for a weather-app coding scenario.
Findings:
- Model Performance: Solid, with good tool use and overall output quality.
- Usability: Easy to use local models within the Copilot app.
- Bottleneck: Hardware limits remain real for local inference when balancing speed and quality in agentic tasks.
While the model is capable, running high-load agent tasks locally smoothly still demands more powerful hardware.
More from coding & agent
- Cyclomatic complexity audit cuts decision paths from 91 to 12 — DanielLockyer · 2026-08-27
- ChatGPT Adds Skills-over-MCP Support for Synced Agent Workflows — iamrobotbear · 2026-08-27
- Developer gives Claude a domain and lets it build whatever it wants — United-Combination66 · 2026-08-27
- Self-audit of a memory MCP server found models could read other users' memories — Technical_Bench_188 · 2026-08-27
- Is Document Parsing the Real Bottleneck in Your RAG System? — -R-I-k- · 2026-08-27
- OMEM: fully local agent memory layer that doesn't use a model to decide truth — Technical_Bench_188 · 2026-08-27