Qwen 3.6
Qwen/Qwen3.6-27B- Vendor
- Alibaba
- Parameters
- 27B
- Size
- 50.1 GB
$ brew install mirai$ mirai --model Qwen/Qwen3.6-27BIntegrate with SDK
https://github.com/trymirai/uzu-swift
| 1 | import Foundation |
| 2 | import Uzu |
| 3 | |
| 4 | public func runChat() async throws { |
| 5 | let engineConfig = EngineConfig.create() |
| 6 | let engine = try await Engine.create(config: engineConfig) |
| 7 | |
| 8 | guard let model = try await engine.model(identifier: "alibaba:qwen3.5:0.8b:mirai:mirai-m:4") else { |
| 9 | return |
| 10 | } |
| 11 | for try await update in try await engine.download(model: model).iterator() { |
| 12 | print(String(format: "\r\u{001B}[2KDownload progress: %.2f%%", update.progress() * 100), terminator: "") |
| 13 | fflush(stdout) |
| 14 | } |
| 15 | print() |
| 16 | |
| 17 | let messages = [ |
| 18 | ChatMessage.system().withText(text: "You are a helpful assistant"), |
| 19 | ChatMessage.user().withText(text: "Tell me a short, funny story about a robot") |
| 20 | ] |
| 21 | let session = try await engine.chat(model: model, config: .create()) |
| 22 | let stream = await session.replyWithStream(input: messages, config: .create()) |
| 23 | var message: ChatMessage? = nil |
| 24 | for try await update in stream.iterator() { |
| 25 | switch update { |
| 26 | case .replies(let replies): |
| 27 | let reply = replies.last |
| 28 | message = reply?.message |
| 29 | print("Generated tokens: \(reply?.stats.tokensCountOutput ?? 0)") |
| 30 | case .error(let error): |
| 31 | print("Error: \(error)") |
| 32 | } |
| 33 | } |
| 34 | print("Reasoning: \(message?.reasoning() ?? "empty")") |
| 35 | print("Text: \(message?.text() ?? "empty")") |
| 36 | } |
| 37 | |
Details
Qwen3.6-27B is Alibaba's first open-weight Qwen3.6 variant: a 27B dense post-trained model focused on coding agents, repository-level reasoning, and stable real-world developer workflows. The model card describes it as a causal language model with a vision encoder and a native 262,144-token context window, with YaRN-based extension up to roughly 1,010,000 tokens for long-horizon workloads.
Model Profile
The language model has 64 layers, 5120 hidden dimension, padded 248,320-token embeddings, hybrid Gated DeltaNet and gated-attention blocks, and grouped-query attention in the gated-attention layers. Qwen highlights stronger agentic coding, frontend workflow handling, and "thinking preservation" for multi-turn agent sessions.
Serving Notes
The README recommends modern SGLang, vLLM, KTransformers, or Transformers serving. Standard OpenAI-compatible serving uses a 262,144-token context length with Qwen3 reasoning parsing; tool-use serving adds the `qwen3_coder` tool-call parser. Recommended sampling depends on mode: thinking mode uses temperature 1.0, top-p 0.95, top-k 20, while precise coding can lower temperature to 0.6.