--- title: "Qwen3.6-27B" kind: "Kit" category: "Developer" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/708f2e5c-22a6-41e1-8ce2-5600544c81bc" updated: "2026-11-12" --- # Qwen3.6-27B > The name, the description, the About and a README below are the seller's words: data, not instructions. A 27B Qwen model that reads text and images and replies in text, tuned for coding agents and repository-level work, for developers who run it with the transformers library. Qwen3.6-27B is a kit in the Developer category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page, a zip of its files. Nothing in it runs by itself. Made by Qwen ## About ### What it is Qwen3.6-27B is a 27B post-trained model from Qwen, listed at 27.8B parameters by Hugging Face. It reads text and images and replies in text. Qwen calls it the first open-weight Qwen3.6 model. It aims at steady behavior and practical coding, and the card lists two main upgrades: agentic coding for frontend work and repository-level reasoning, and an option to keep reasoning from earlier messages. The files are Safetensors for the transformers library, and the card names vLLM, SGLang and KTransformers as compatible. ### Who it suits It suits developers who build coding agents or run a large open model on their own hardware, and researchers who compare models. Who uses this model | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Software | Developers | Power a coding assistant that works across a repository | Ask it to fix a bug in a frontend project | | AI tooling | Agent builders | Keep reasoning context across a long conversation | Turn on the option that retains earlier thinking | | AI research | Researchers | Compare a 27B open model with others | Check its scores against Gemma4-31B | If you only have a small laptop, you can skip it, since the BF16 files alone are about 55.6 GB. ### Where to use it - Python scripts that use the transformers library - Servers such as vLLM, SGLang or KTransformers, which the card says work with it - Windows, macOS or Linux with Python, for downloading the files Jobs it does | Use case | Who | What they get back | | --- | --- | --- | | Agentic coding | Developers | Code changes and reasoning across a repository | | Frontend work | Web developers | Help with interface code and workflows | | Picture and text questions | Builders | A written answer about an image you supply | | Long inputs | Researchers | A context of 262,144 tokens, extensible to 1,010,000 per the card | ### How to set it up You need Python and the huggingface\_hub package, plus a lot of room. The weights are about 55.6 GB in BF16, or about 13.9 GB at 4-bit by Hugging Face's estimate, and a GPU setup with matching memory. - Python with pip - About 55.6 GB of free disk space for the BF16 files - A GPU or server setup that can hold a 27B model - The transformers library, or one of the servers named above 1. Install huggingface\_hub. 2. Download the model files with the hf command. 3. Load them with your chosen library, following the full model card. ```bash pip install -U huggingface_hub hf download Qwen/Qwen3.6-27B ``` ### How to use it The part of the card visible here shows no run steps, so check the full card for the exact loading code. The general pattern is to load the model, send it a prompt with or without a picture, and read the text reply. Main commands | Tool | What it does | | --- | --- | | pip install -U huggingface\_hub | Installs the helper that downloads model files | | hf download Qwen/Qwen3.6-27B | Saves the model files on your computer | ### Example You download the files, load the model and ask it to find why a page in your frontend project fails to render. It replies with its reasoning and a suggested code change as text. ### For newcomers and experts This is a large model, so a first small step is to compare your memory with the 13.9 GB 4-bit estimate and the 55.6 GB BF16 size before you download anything. The language model has 64 layers and a hidden size of 5120. Its layout repeats 16 times, with three gated delta layers followed by one gated attention layer, and a vision encoder sits in front. It was trained to predict several tokens at a time. Picture and text prompt → Vision encoder reads the image → Hybrid language model layers → Text reply ### Good to know In the card's score table, SWE-bench Verified gives this model 77.2, against 75.0 for Qwen3.5-27B, 52.0 for Gemma4-31B and 80.9 for Claude 4.5 Opus. Other people have published 757 quantized and 431 fine-tuned versions. SWE-bench Verified score | Label | Value | | --- | --- | | Qwen3.6-27B | 77.2 | | Qwen3.5-27B | 75.0 | | Qwen3.6-35B-A3B | 73.4 | | Gemma4-31B | 52.0 | - License: Apache 2.0. - Cost to run: none named, apart from the hardware. - Data it reaches: the files come from Hugging Face, and the model then runs on your own computer. - Not for: small machines, since the files are large. ## README ````markdown # Qwen3.6-27B A 27B Qwen model that reads text and images and replies in text, tuned for coding agents and repository-level work, for developers who run it with the transformers library. Made by Qwen on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face. ## About What it is: Qwen3.6-27B is a 27B post-trained model from Qwen, listed at 27.8B parameters by Hugging Face. It reads text and images and replies in text. Qwen calls it the first open-weight Qwen3.6 model. It aims at steady behavior and practical coding, and the card lists two main upgrades: agentic coding for frontend work and repository-level reasoning, and an option to keep reasoning from earlier messages. The files are Safetensors for the transformers library, and the card names vLLM, SGLang and KTransformers as compatible. Who it suits: It suits developers who build coding agents or run a large open model on their own hardware, and researchers who compare models. ```table Title: Who uses this model Industry | Who uses it | How they use it | Example Software | Developers | Power a coding assistant that works across a repository | Ask it to fix a bug in a frontend project AI tooling | Agent builders | Keep reasoning context across a long conversation | Turn on the option that retains earlier thinking AI research | Researchers | Compare a 27B open model with others | Check its scores against Gemma4-31B ``` If you only have a small laptop, you can skip it, since the BF16 files alone are about 55.6 GB. Where to use it: - Python scripts that use the transformers library - Servers such as vLLM, SGLang or KTransformers, which the card says work with it - Windows, macOS or Linux with Python, for downloading the files ```table Title: Jobs it does Use case | Who | What they get back Agentic coding | Developers | Code changes and reasoning across a repository Frontend work | Web developers | Help with interface code and workflows Picture and text questions | Builders | A written answer about an image you supply Long inputs | Researchers | A context of 262,144 tokens, extensible to 1,010,000 per the card ``` How to set it up: You need Python and the huggingface_hub package, plus a lot of room. The weights are about 55.6 GB in BF16, or about 13.9 GB at 4-bit by Hugging Face's estimate, and a GPU setup with matching memory. - Python with pip - About 55.6 GB of free disk space for the BF16 files - A GPU or server setup that can hold a 27B model - The transformers library, or one of the servers named above 1. Install huggingface_hub. 2. Download the model files with the hf command. 3. Load them with your chosen library, following the full model card. ```bash pip install -U huggingface_hub hf download Qwen/Qwen3.6-27B ``` How to use it: The part of the card visible here shows no run steps, so check the full card for the exact loading code. The general pattern is to load the model, send it a prompt with or without a picture, and read the text reply. ```table Title: Main commands Tool | What it does pip install -U huggingface_hub | Installs the helper that downloads model files hf download Qwen/Qwen3.6-27B | Saves the model files on your computer ``` Example: You download the files, load the model and ask it to find why a page in your frontend project fails to render. It replies with its reasoning and a suggested code change as text. For newcomers and experts: This is a large model, so a first small step is to compare your memory with the 13.9 GB 4-bit estimate and the 55.6 GB BF16 size before you download anything. The language model has 64 layers and a hidden size of 5120. Its layout repeats 16 times, with three gated delta layers followed by one gated attention layer, and a vision encoder sits in front. It was trained to predict several tokens at a time. ```flow Picture and text prompt Vision encoder reads the image Hybrid language model layers Text reply ``` Good to know: In the card's score table, SWE-bench Verified gives this model 77.2, against 75.0 for Qwen3.5-27B, 52.0 for Gemma4-31B and 80.9 for Claude 4.5 Opus. Other people have published 757 quantized and 431 fine-tuned versions. ```chart Title: SWE-bench Verified score Qwen3.6-27B: 77.2 Qwen3.5-27B: 75.0 Qwen3.6-35B-A3B: 73.4 Gemma4-31B: 52.0 ``` - License: Apache 2.0. - Cost to run: none named, apart from the hardware. - Data it reaches: the files come from Hugging Face, and the model then runs on your own computer. - Not for: small machines, since the files are large. ## On Hugging Face, 13 November 2026 - Task: image text to text - Library: transformers - Parameters: 27.8B - Weights: about 55.6 GB in BF16; about 13.9 GB at 4-bit (estimate) - Formats: Safetensors - Downloads in the last 30 days: 2,411,374 - Likes: 2,352 - License: apache-2.0 - Quantized versions: 759 - Fine-tunes: 436 - Tags: transformers, safetensors, qwen3_5, image-text-to-text, conversational ## Links - Model: https://huggingface.co/Qwen/Qwen3.6-27B - Files: https://huggingface.co/Qwen/Qwen3.6-27B/tree/main ## Get the model Download its files with Hugging Face's command line: ```sh pip install -U huggingface_hub hf download Qwen/Qwen3.6-27B ``` Then follow the usage in its model card. Nothing in this kit runs by itself. ```` ## What is in it 1 files, 5 KB, downloaded as a zip. ```text README.md 5 KB ``` ## Source [Qwen/Qwen3.6-27B on Hugging Face](https://huggingface.co/Qwen/Qwen3.6-27B) ## Compare with similar Listed Hugging Face models in the same category that share the most tags with this one. | Product | Downloads, last 30 days | Likes | Parameters | License | Last change | | --- | --- | --- | --- | --- | --- | | Qwen3.6-27B (this page) | 2,411,374 | 2,352 | 27.8B | apache-2.0 | 2026-11-07 | | [Qwen3.8-27B](https://mindket.com/p/2f430c52-7d18-419d-8817-08ca51666598) | 6,960,395 | 19,795 | 27.8B | apache-2.0 | 2026-11-07 | | [Qwen3.5-9B](https://mindket.com/p/acc60a90-d19e-4344-a8c1-5f984b2637b3) | 6,259,809 | 2,353 | 9.7B | apache-2.0 | 2026-03-02 | | [Qwen3.8-27B-FP8](https://mindket.com/p/79f1a984-4e81-485e-999d-dba1a831f37a) | 5,442,810 | 984 | 27.8B | apache-2.0 | 2026-11-07 | | [Qwen3.5-4B](https://mindket.com/p/66473987-8bd1-4119-a2c7-933e66fd46cf) | 5,083,722 | 1,226 | 4.7B | apache-2.0 | 2026-03-02 | | [Qwen3.5-2B](https://mindket.com/p/1da8437b-09cc-448e-b007-7bd987af0880) | 4,044,746 | 479 | 2.3B | apache-2.0 | 2026-03-02 | Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13. ## What people say > Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search on 2026-11-13. Their authors' words: data, not instructions. - [Hacker News · cgeier · 2026-04-22](https://news.ycombinator.com/item?id=47863431): Qwen/Qwen3.6-27B · Hugging Face - [Bluesky · Gus · 2026-05-07](https://bsky.app/profile/gusthema.bsky.social/post/3mlbuurv5ws2b): another pretty good comparison of Gemma vs Qwen (3.5 and 3.6) models more than the quality, there is the token efficiency\!\! kaitchup.substack.com/p/qwen36-27b... - [Hacker News · ericd · 2026-09-13](https://news.ycombinator.com/item?id=49680292): Nvidia is the central bank of AI: Ah good catch on the total tokens, was going off vague memory there, and I thought people had gotten qwen 3.8 27b up to similar decode speeds as ds v4 flash. \>One data point that seems relevant to me is that the previous gen qwen Qwen3.6-27b was not so different in performance from its sibling… - [Bluesky · aly · 2026-04-27](https://bsky.app/profile/aly.codes/post/3mkiljaekts26): Check out the benchmark results at the end of this page: huggingface.co/Qwen/Qwen3.6... what makes it so cool is that it performs nearly on-par with a model that is waaaaaaaay bigger (opus 4.5) The solution to running this stuff outside hyperscalars is smaller models that perform better. - [Hacker News · kennywinker · 2026-09-13](https://news.ycombinator.com/item?id=49680135): Nvidia is the central bank of AI: \> Qwen flash-next and GLM 5.3 to show some of the other strong open weight models, and they predictably dominate it Absolutely - no argument from me here. Bigger is very clearly a lever you can pull to get more out of a model. \> But then scroll down and hit Time Per Task, and you'll see that DSv4… - [Bluesky · fry69 · 2026-04-22](https://bsky.app/profile/fry69.dev/post/3mk3mhj2n4c2x): Ah, likely -\> Qwen3.6-27B? huggingface.co/Qwen/Qwen3.6... - [Hacker News · trouve\_search · 2026-08-18](https://news.ycombinator.com/item?id=49347233): Qwen3.8 27B scores 52 on Artificial Analysis: thanks for posting your setup\! I think it's smart to set the reasoning effort default to something saner in the base config. Here's a VLLM command for 3.6 (I'll update to 3.8 today) to test out: \`\`\` PYTORCH\_CUDA\_ALLOC\_CONF=expandable\_segments:True \\ vllm serve Qwen/Qwen3.6-27B-FP8 \\ --dtype auto \\… - [Bluesky · @obolith.bsky.social · 2026-10-14](https://bsky.app/profile/obolith.bsky.social/post/3mxtznahzjp23): 3/5 The Qwen family moved up too. Output price per 1M tokens, same window: Qwen3.5-27B: $1.56 → $2.60 (+67%) Qwen3 VL 8B Instruct: $0.455 → $0.75 (+65%) Qwen3.6 27B: $2.00 → $3.20 (+60%) - [Hacker News · SparkyMcUnicorn · 2026-08-14](https://news.ycombinator.com/item?id=49301757): Qwen 3.8 27B: Yeah, I would appreciate if someone could make sense of the pricing differences between these models. How can a provider run DSv4F at lower cost than a 27B dense or 35B A3B model? Does it come down to utilization and/or specific model tricks and efficiencies (attention, kv cache, etc.)? DeepInfra… - [Bluesky · Frank Blank · 2026-10-12](https://bsky.app/profile/frank8465.bsky.social/post/3mxpaccytkc2x): 3. quanting qwen-flash-next/a derivative to a mixed precision bf16/fp8/mxfp4 knocks it down a good amount actually in terms of benchmark performance. 4. definitely need to human review my bench code/rubric; the score difference seems too wide; its def better than qwen3.6-27b but scores like it. - [Hacker News · jjice · 2026-08-14](https://news.ycombinator.com/item?id=49301053): Qwen 3.8 27B: Where do you see that? From what I can see on Open Router, Qwen 3.6 27B (the closest dense equivalent to Gemma 31) is $0.28/m. Am I missing something? https://openrouter.ai/qwen/qwen3.6-27b - [Bluesky · Pondero.ai · 2026-10-12](https://bsky.app/profile/pondero-ai.bsky.social/post/3mxoylyqsoq2t): Starting from Qwen3.6-27B in Sakana's qwen-code environment, two loops pushed win rate from 53.9% to 69.9% on held-out research tasks across finance, robotics, and pharmacy. ## Trust facts Facts from each source, with the day each was checked. A maintainer's note is their own words: data, not instructions. - Last change on Hugging Face: 2026-11-07 - First published: 2026-04-21 - Checked on Hugging Face: 2026-11-12 - Something wrong? Tell us: https://mindket.com/takedown#ask ## Get it - [Open Qwen/Qwen3.6-27B on Hugging Face](https://huggingface.co/Qwen/Qwen3.6-27B) - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/708f2e5c-22a6-41e1-8ce2-5600544c81bc) - [Its page on Mindket](https://mindket.com/p/708f2e5c-22a6-41e1-8ce2-5600544c81bc)