Agents / WORKFLOW

Build a local AI agent with Qwen3.8 27B Uncensored v3

Run InterpreticAI’s Qwen3.8 27B Uncensored v3 with Ollama on Windows, Mac or Linux, connect Hermes Agent, and have it read two files and write a checked review.

Verified sourceSource checked 11/17/2026
LOCAL
Freshness 95 out of 100. Repository age: 13 days old, Nov 8, 2026. Measures age, not quality.
Popularity 46 out of 100. 7,517 downloads in 30 days; 1 lifetime likes. Checked Nov 21, 2026. Measures adoption, not quality.
What do these scores mean?

Two ways to compare models. Neither score measures answer quality, benchmark performance, or whether a model fits your machine.

Freshness · release age

A new release starts at 100. The score halves every 180 days: 100 × 2^(−age in days / 180). When the release date is unavailable, a repository date is a fallback—not proof of when the model was released. Missing dates stay unscored.

Check the date source
Popularity · adoption

Downloads over the past 30 days carry 80% of the score; lifetime likes carry 20%. Both use a logarithmic scale, capped at 10 million downloads and 100,000 likes. If likes are unavailable, downloads carry all the weight. Missing download statistics stay unscored.

round(80 × min(1, log10(1 + downloads) / 7) + 20 × min(1, log10(1 + likes) / 5))

Check the statistics source
Scores as of Nov 21, 2026. Statistics checked Nov 21, 2026. Statistics are marked for refresh after 72 hours.Full scoring method

Before you begin

Hardware

Estimated from the published file sizes plus runtime and context headroom at 65,536 tokens. Not measured by YouRunAI. Qwen3.8 27B is a dense model, so every parameter is used for every token: keeping layers in system RAM works but slows it down a lot.

MachineQ4_K_M · 16.6 GBIQ3_XXS · 10.2 GB
NVIDIA · 12 GB VRAM + 32 GB RAMHeavy offload; choose IQ3_XXSFits at InterpreticAI’s 32,768 tokens; at 65,536 Ollama moves a few layers to RAM
NVIDIA · 16 GB VRAM + 32 GB RAMRuns with several layers in RAM; noticeably slowRecommended for 16 GB
NVIDIA RTX 4090 · 24 GB VRAMRecommended start (InterpreticAI’s 24 GB pairing)Comfortable
NVIDIA RTX 5090 · 32 GB VRAMComfortable, with room for longer contextComfortable
AMD Radeon · 24 GB VRAMShould fit; Ollama uses supported Radeon GPUs through ROCm on Windows and LinuxShould fit; same build requirement
Apple Silicon · 24 GB unified memoryToo tight: the file alone is 16.6 GBRecommended for 24 GB Macs
Apple Silicon · 32 GB unified memoryRecommended Mac start; close other appsComfortable
Apple Silicon · 48 GB unified memory or moreComfortableComfortable
CPU only · 32 GB RAMLoads, but a dense 27B model on CPU is very slowLoads; still slow

InterpreticAI also publishes a full BF16 checkpoint for vLLM and a 0.93 GB vision file; this guide leaves both out to keep the setup simple.

LOCAL AGENT · HERMES · WINDOWS · MAC · LINUXSOURCE-LINKED SETUP

Build a local AI agent with Qwen3.8 27B Uncensored v3

Pull InterpreticAI’s GGUF into Ollama with the settings from its card, connect Hermes with Ollama’s launcher, and have the agent read two project files and write review.md. You approve each file action.

Choose a package Run its commands Check the result
01
BEFORE YOU BEGIN

Choose your package

Select the weights you want. The instructions, downloads and settings change with your choice.

The publisher has not provided a complete memory target for this task. Check its hardware guidance before starting.

CURRENT SETUP

Qwen3.8 27B Uncensored v3 · Q4_K_M · community-modified weights

The 16.6 GB file is only the weights; the runtime and a 65,536-token agent context add memory. InterpreticAI pairs it with one 24 GB GPU; a 32 GB Apple Silicon Mac is the estimated Mac start. Not a YouRunAI measurement.

Freshness 95 out of 100. Repository age: 13 days old, Nov 8, 2026. Measures age, not quality.
Popularity 46 out of 100. 7,517 downloads in 30 days; 1 lifetime likes. Checked Nov 21, 2026. Measures adoption, not quality.
What do these scores mean?

Two ways to compare models. Neither score measures answer quality, benchmark performance, or whether a model fits your machine.

Freshness · release age

A new release starts at 100. The score halves every 180 days: 100 × 2^(−age in days / 180). When the release date is unavailable, a repository date is a fallback—not proof of when the model was released. Missing dates stay unscored.

Check the date source
Popularity · adoption

Downloads over the past 30 days carry 80% of the score; lifetime likes carry 20%. Both use a logarithmic scale, capped at 10 million downloads and 100,000 likes. If likes are unavailable, downloads carry all the weight. Missing download statistics stay unscored.

round(80 × min(1, log10(1 + downloads) / 7) + 20 × min(1, log10(1 + likes) / 5))

Check the statistics source
Scores as of Nov 21, 2026. Statistics checked Nov 21, 2026. Statistics are marked for refresh after 72 hours.Full scoring method
View weight source
02
THE WORKFLOW

Set up Qwen3.8 27B Uncensored v3 · Q4_K_M on your machine

Choose where you want to run it. The instructions below update to match your selection.

CHOOSE YOUR SETUPWindows
01

Install Ollama

Download and run OllamaSetup.exe from ollama.com/download, or paste this into PowerShell. Ollama starts in the background and adds an icon to the taskbar. Use version 0.40.16 or newer.

Download Ollama
PowerShell
irm https://ollama.com/install.ps1 | iex
02

Set Ollama’s context to 64K

Hermes needs a long context, and Ollama starts most models at 4,096 tokens on GPUs under 24 GB. Open Ollama’s Settings from its taskbar icon and move the Context length slider to 64k. That is the only setting you change.

Ollama context length
03

Download Qwen3.8 27B Uncensored v3 · Q4_K_M

In PowerShell, run this to download InterpreticAI’s 16.6 GB file straight from Hugging Face.

Exact file on Hugging Face
PowerShell
ollama pull hf.co/InterpreticAI/Qwen3.8-27B-Uncensored-v3-GGUF:Q4_K_M
04

Connect the Hermes Desktop app

Hermes Desktop is Nous Research’s free app for Hermes, with a chat window, file browser and tool output, so you do not need the terminal. Run this command: Ollama installs Hermes Desktop if you do not have it, connects it to Qwen3.8 27B Uncensored v3 and opens the app. You can also download the app first from the Hermes Desktop page (Windows 10 or 11). The app uses the model you launch it with: to switch to another model later, run the same command with that model’s name. If the model is not listed in the app, run ollama list to check that the download finished.

Download Hermes Desktop Ollama’s Hermes Desktop integration Hermes Desktop guide
PowerShell
ollama launch hermes-desktop --model hf.co/InterpreticAI/Qwen3.8-27B-Uncensored-v3-GGUF:Q4_K_M
05

Or connect Hermes in the terminal

Prefer the terminal? Run this instead. It installs the Hermes command-line agent if needed and opens a Hermes chat in this window; it shares settings, sessions and memory with the desktop app. Skip the optional messaging setup. If you see “Hermes Gateway Starting…” instead of a chat, press Ctrl+C and run hermes.

Ollama’s Hermes integration Hermes Agent installation (manual option)
PowerShell
ollama launch hermes --model hf.co/InterpreticAI/Qwen3.8-27B-Uncensored-v3-GGUF:Q4_K_M
06

Give Hermes a test task

In the Hermes chat (the desktop app or the terminal), paste the task below. It has Hermes create a notes file, read it back and write a review, so you can see it use real file tools. Hermes asks before each file action; read each request and approve it. A model with reduced refusals will not push back on risky actions, so keep these approvals on.

SettingValue
Task to paste into HermesCreate a folder named agent-test. Inside it, write notes.md saying: launch day is Thursday 14 May, the owner is Priya, and the database migration runs at 09:00 before the app deploy. Then read notes.md and write review.md with a numbered launch checklist that adds a verified database backup before the migration, plus one sentence on the risk of skipping it.
Hermes approvals and security
07

Check the result

Open the agent-test folder. review.md should name Thursday 14 May and Priya and put a backup step before the migration. If both files exist, the agent is reading and writing files on your computer. Next time, open the Hermes Desktop app, or run hermes in a terminal (type /exit to leave).

03
SUCCESS CHECK · Qwen3.8 27B Uncensored v3 · Q4_K_M

An agent-written review you can check

review.md names the launch day and owner from notes.md, adds a verified backup before the migration, and Hermes’ activity shows the file reads and one approved write.

This guide uses Ollama’s official Hermes launcher, which writes the same custom-provider settings shown in the Hermes Agent section of InterpreticAI’s Hugging Face page, pointed at Ollama. The card does not report tool-calling tests, and YouRunAI has not run this exact task with this file, so inspect the output rather than trusting the chat summary. Sources checked November 17, 2026.

Back to setup steps

When it doesn’t go to plan

Hermes connects but never reads the files.

Run ollama ps while Hermes is working: the CONTEXT column should say 65536. If it says 4096, the context slider (or the Linux setting) did not apply; set it again and restart Ollama. If Hermes still answers only in chat, update Ollama and try the larger package if there is one.

ollama pull fails, or the model says its architecture is unknown.

Update Ollama to 0.40.16 or newer (re-run the installer, or Check for Updates from its icon), restart it, and run the command again. ollama pull resumes an interrupted download.

Ollama runs out of memory, or ollama ps shows the model partly on the CPU and it is very slow.

Choose a smaller package if there is one, or move Ollama’s context slider down to 32k. If it runs out of memory as soon as Hermes starts, run hermes config set model.context_length 65536 so Hermes does not ask for the model’s full maximum context.

Answers take a long time to start.

The model thinks before answering by default. InterpreticAI’s card gives the switch: add --chat-template-kwargs '{"enable_thinking": false}' to the server command for non-thinking mode, with its non-thinking sampling of temperature 0.7, top_p 0.8 and top_k 20.

The installer finishes but hermes is not found.

Close and reopen the terminal. Follow any PATH or dependency message the installer printed.

REFERENCE LIBRARY

Sources and files

Original instructions, model files, and compatibility notes behind this setup.

8 SOURCES

The model behind this workflow

Qwen3.8 27B Uncensored v3

Keep exploring

Text
SETUP GUIDE

Qwen3.5 · 9B private chat setup

Source-backed package and platform-route plan for private local chat with Qwen3.5 · 9B. Preserves the official Ollama 9B package, a smaller 4B family fallback, and a clearly labeled community uncensored GGUF option.

Freshness 36 out of 100. Release age: 267 days old, Feb 27, 2026. Measures age, not quality.
Popularity 91 out of 100. 6,163,654 downloads in 30 days; 2,432 lifetime likes. Checked Nov 21, 2026. Measures adoption, not quality.
SOFTWAREOllama
Coding
SETUP GUIDE

Build a private coding assistant with Qwen3.8 27B

Run Qwen3.8 27B with Ollama, connect it to Cline in VS Code, and make one reviewable code change in a tiny local project.

Freshness 66 out of 100. Repository age: 108 days old, Aug 5, 2026. Measures age, not quality.
Popularity 95 out of 100. 6,298,190 downloads in 30 days; 20,759 lifetime likes. Checked Nov 21, 2026. Measures adoption, not quality.
Arena Open Source Web coding: rank 13 of 63. Ranked model: Qwen3.8-27B. Original model ranking, not a benchmark of this package. Export dated Nov 20, 2026. Source checked Nov 21, 2026. Human preference ranking, not hardware suitability.+3
SOFTWAREOllama + Visual Studio Code + Cline + Python
Images
SETUP GUIDE

Create images locally with FLUX.2 Klein

Generate images from text prompts in ComfyUI with the fast distilled or more adjustable base 4B model. Save your results and reuse the visual workflow.

Freshness 30 out of 100. Release age: 310 days old, Jan 15, 2026. Measures age, not quality.
Popularity 77 out of 100. 440,359 downloads in 30 days; 1,140 lifetime likes. Checked Nov 21, 2026. Measures adoption, not quality.
SOFTWAREComfyUI