Distill Opus-level quality into Qwen 3.8 27B with a LoRA and drop your API bill
alexcovo_eth · x · 2026-11-11
SimonHoiberg's team had a frontier model work through thousands of realistic support cases, kept only the outputs meeting their quality bar, and trained a LoRA on Qwen 3.8 27B with that data. The distilled small model reportedly matches Opus 5.5-level quality on the task while running on their own hardware — no API bill, no rate limits, no risk of the vendor changing the deal.
More from coding & agent
- Google wires CodeMender into pull requests: /fix comments auto-patch vulnerabilities — rseroter · 2026-11-11
- Dev builds a private message board so his AI agents can share one machine — D3VAUX · 2026-11-11
- How should UIs attribute actions taken by agents through a signed-in human account? — max__drake · 2026-11-11
- Fable 5.2 audits an Opus 5.5-built codebase and finds issues Opus missed — CtrlAltDwayne · 2026-11-11
- Google Antigravity ships Swarm Mode: parallel agents coordinating via shared message board — testingcatalog · 2026-11-11
- Comet Builds Docker-Based Incident Testbed to Evaluate Remediation Agents with Opik — craigsdennis · 2026-11-11