Qwen 3.8 27B: what developers actually say

We read every Hacker News comment from Nov 2025 – Nov 2026 that gave a real opinion on Qwen 3.8 27B. 55 of the 70 opinions come from people who say they use it themselves.

69%approval, all opinions

69%approval, last 90 days

#23of 45 models ranked

59different developers

Qwen 3.8 27B has 69% approval across 70 opinions from 59 authors, with 55 opinions firsthand. Approval rose from 57% in August to 72% in September and 82% in October; November has one opinion at 100%, so that month is a sample of one. Code quality gets 16 positive and 4 negative comments, privacy/local 12 positive and none negative, and autonomy 10 positive and 1 negative. Speed gets 9 positive and 5 negative comments, and reliability gets 1 positive and 4 negative, so fast, predictable edits are the weak point.

Pick it if

Want local or private inference (privacy/local: 12 positive, 0 negative)
Run long or overnight agent tasks (autonomy: 10 positive, 1 negative)
Accept slower reasoning for 27B code output (code quality: 16 positive, 4 negative)

Look elsewhere if

Need fast responses for interactive work (speed: 9 positive, 5 negative)
Need dependable edit application in local coding tools (reliability: 1 positive, 4 negative)
Need complex implementation without long churn (users report overthinking and hours of churn)

Strengths and weaknesses

Each row counts opinions that mentioned the topic. Bars to the right of the line are praise, bars to the left are complaints.

PraiseComplaints

Code quality

Speed

Privacy & local use

Autonomy

Reliability

How opinion has moved

Monthly approval (line) and number of opinions (bars). Months with fewer than 3 opinions are left out.

What developers say

Praise

Says the Q6 dense 27B model fixed most of the bugs that Flash Next introduced at low quantization.HN comment · Nov 1, 2026

Says Qwen 3.8 27B at Q3 on their own AMD GPU works well enough for tool calling in agents.HN comment · Oct 30, 2026

Runs Qwen 3.8 27B locally at about 50 tokens/sec and calls it fine for background agentic tasks.HN comment · Oct 25, 2026

Says Qwen 3.8 27B is the only model they use mostly, running agentic planning workflows with minimal harness.HN comment · Oct 25, 2026

Says a Qwen 3.8 27B model they post-trained completed an end-to-end PHP exploit through their agent setup.HN comment · Oct 22, 2026

Own benchmarks show Qwen 3.8 27B beats Haiku and Sonnet 4.5 on coding, and runs fast locally.HN comment · Oct 19, 2026

Says Qwen3.8-27B keeps its agent loop running until it claims success, so it's useful for overnight coding runs.HN comment · Oct 15, 2026

Says Qwen 3.8 27B is far better than earlier local models and does good coding for its size.HN comment · Oct 3, 2026

Complaints

Says the Thinking Cap and Swift variants of Qwen 3.8 27B were less capable than the base model on their tasks.HN comment · Oct 23, 2026

Says Qwen 3.8 27B runs fast locally but is slow to reach reviewed results in a multi-model setup.HN comment · Sep 24, 2026

Says Qwen3.8-27B is too slow on a local M1 Max even with MTP acceleration.HN comment · Sep 16, 2026

Says an abliterated Qwen 3.8 27B ignores prompt restrictions, so rules must be enforced in the harness or sandbox.HN comment · Sep 13, 2026

Reports Qwen 3.8 27B overthinks, churns for hours on a task, and is not good enough for complex implementation.HN comment · Sep 9, 2026

Says Qwen 3.8 27B is usable but slow to reason.HN comment · Sep 1, 2026

Says Qwen 3.8 27B run locally in Claude Code and OpenCode fails edits far more often than cloud models.HN comment · Aug 31, 2026

Found Qwen 3.8 27B useless for anything non-trivial despite days of trying.HN comment · Aug 26, 2026

Summaries are paraphrased. Each links to the original comment.

FAQ

Is Qwen 3.8 27B worth it?

For local and background coding work, often yes. It has 69% approval, with strong feedback on privacy and autonomy. For fast, reliable interactive coding, the feedback is weaker.

Is Qwen 3.8 27B better than Claude Sonnet 4.5?

The data is too thin to say. One user's benchmark puts it ahead on coding, and one person preferred it over Sonnet 4.5. Other users report more failed edits than with cloud models.

Can Qwen 3.8 27B run locally?

Yes, and local use is its strongest theme: 12 positive comments and no negative ones on privacy/local. Users report running it at Q3 on an AMD GPU and at about 50 tokens per second. One M1 Max user still finds it too slow.