Local 27B writes all the code, cloud model just plans: matches frontier quality at ~10% cost
GapNew4766 · reddit · 2026-11-19
A Reddit user tested a planner/worker split where cloud qwen4-max only plans and reviews while a local Qwen 27B (q4 quant, llama.cpp on a 64GB M5 Max) writes every line of code.
Setup
- Task: single-file HTML canvas game with 3 levels, arrow keys, and a ?demo=1 self-playing mode
- Grading: headless Chrome runs 10 checks (no console errors, keys work, demo scores, reaches level 2/3 and end screen)
- Same prompt, temp 0.2, 20-min limit, 2 runs each
Results
- qwen4-max alone: 10/10 both runs, 6/15 min, costing 13–21 cents
- Local 27B alone: 2/10 and 8/10 — it discarded correct fixes, couldn't find its own syntax error, and looped on failing checks until timeout
- Fusion split: 10/10 both runs at just 1–9 cents; in run 2 the 27B hit a syntax error but fixed it on rewrite instead of getting stuck, thanks to the upfront plan
Takeaway: planning fully closed the gap for the small local model while cutting cost 90%, with all code written locally. Only 2 runs per setup, so a strong signal rather than proof.
Related event: Planner-Worker Architecture Cuts Coding Costs by 90% with Local 27B Model(2 posts)→
More from coding & agent
- ClaudeAI Weekly: Single-HTML-File Arms Race, 8.9B-Token Strawberry, Post-IPO Anthropic — ClaudeAI-mod-bot · 2026-11-22
- Agno's AgentOS turns any agent into a production FastAPI app with MCP, RBAC and tracing — pritisinghhhh · 2026-11-22
- Chinese devs vent: app store rejections and copycats are just the norm — tinyfool · 2026-11-22
- Compacting Early and Often Boosts Agent Accuracy While Cutting Cost and Runtime — zainhas · 2026-11-22
- Astra 6.2 silently skips modules in audit tasks, then claims it checked everything — MbahSurip · 2026-11-22
- Is Haskell 'unsafe' too? A debate on types vs verification in the LLM era — satnam6502 · 2026-11-22