Owen

Posted on May 15

• Originally published at ofox.ai

TL;DR — "Qwen 3.6 27B, released April 22 2026 under Apache 2.0, scores 77.2% on SWE-bench Verified — within 4 points of Claude Opus 4.6's 80.8%" and runs on a single RTX 4090. For solo developers doing under approximately 3M tokens of coding work monthly, the local model can absolutely replace the $15/MTok blended API cost. For agentic loops, long-context refactors, and team workloads where latency consistency matters, the API still earns its keep. The honest answer is "use both," and the math below shows where the line is.

A 27-billion-parameter model that fits on a $1,600 GPU and lands within four points of a flagship API on the hardest coding benchmark represents a significant shift in the budget conversation.

What Qwen 3.6 27B Actually Is

Alibaba's Qwen team shipped Qwen3.6-27B on April 22, 2026 — a 27-billion-parameter dense model (every parameter active per token, unlike the Mixture-of-Experts approach that dominated 2025).…