Workflows/Build a local AI agent with Qwen3.8 Flash Next Uncensored
Agents / WORKFLOW
Build a local AI agent with Qwen3.8 Flash Next Uncensored
Run Qwen3.8 Flash Next Uncensored with llama.cpp on a 128 GB Windows, Mac or Linux machine, connect Hermes Agent, and have it read two files and write a checked review.
Verified sourceSource checked 11/17/2026
LOCAL
Freshness 72 out of 100. Repository age: 87 days old, Aug 26, 2026. Measures age, not quality.72/ 100Freshness87 days oldRepository age · Aug 26, 2026
Popularity 79 out of 100. 673,135 downloads in 30 days; 1,551 lifetime likes. Checked Nov 21, 2026. Measures adoption, not quality.79/ 100Popularity673.1K downloads / 30d1.6K lifetime likes · Checked Nov 21, 2026
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.
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.
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 Flash Next is a mixture-of-experts model, so keeping part of it in system RAM costs less speed than it would with a dense model.
Machine
Qwen3.8 Flash Next Uncensored · IQ4_XS · 98.5 GB
Apple Silicon · 128 GB unified memory
Tight: the file alone is 98.5 GB; close everything else
Apple Silicon · 192 GB unified memory or more
Comfortable
Apple Silicon · 64 or 96 GB unified memory
Cannot fit
PC · one 24–48 GB NVIDIA GPU + 128 GB RAM
Lower -ngl so most of the model sits in RAM; slow
NVIDIA · 141 GB VRAM or more
Comfortable
CPU only · 128 GB RAM
Loads; very slow
NVIDIA DGX Spark is not covered by the Linux tab here, which uses the x86 CUDA Docker image. OrcaRouter’s own GGUF repository (gated) offers other sizes from about 52 GB to 188 GB, and mradermacher lists Q2_K to Q8_0.
LOCAL AGENT · HERMES · WINDOWS · MAC · LINUXSOURCE-LINKED SETUP
Build a local AI agent with Qwen3.8 Flash Next Uncensored
Serve the 98.5 GB IQ4_XS GGUF in llama.cpp, point Hermes at it, and have the agent read two project files and write review.md. You approve each file action. This is a 177B model: plan for a long download and a machine with 128 GB of memory.
Choose a package Run its commands Check the result
01
BEFORE YOU BEGIN
The documented package
This setup uses the package documented for this task. Review its source and supported platforms before starting.
The publisher has not provided a complete memory target for this task. Check its hardware guidance before starting.
CURRENT SETUP
Qwen3.8 Flash Next Uncensored · IQ4_XS · community-modified weights
The 98.5 GB file is only the weights. OrcaRouter says to budget roughly the file size plus the context cache, so a 128 GB machine is tight at a 65,536-token context: close everything else. Not a YouRunAI measurement.
Freshness 72 out of 100. Repository age: 87 days old, Aug 26, 2026. Measures age, not quality.72/ 100Freshness87 days oldRepository age · Aug 26, 2026
Popularity 79 out of 100. 673,135 downloads in 30 days; 1,551 lifetime likes. Checked Nov 21, 2026. Measures adoption, not quality.79/ 100Popularity673.1K downloads / 30d1.6K lifetime likes · Checked Nov 21, 2026
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.
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.
Set up Qwen3.8 Flash Next Uncensored · IQ4_XS on your machine
Choose your computer first, then its GPU path. The instructions below update to match your selection.
CHOOSE YOUR SETUPWindows
01
Install llama.cpp with winget
Open PowerShell and install the official llama.cpp release with winget, then close and reopen PowerShell so llama-server is on your PATH. OrcaRouter’s card says this qwen4_exp architecture needs a llama.cpp build from August 27, 2026 or later; older runtimes report an unknown architecture. The second command prints your build number, and the third lists the GPUs llama.cpp can use. If only the CPU appears on an NVIDIA or AMD PC, download the CUDA or HIP Windows build from the official releases page instead and run the same commands from its folder.
Download Qwen3.8 Flash Next Uncensored · IQ4_XS and start the local server
In PowerShell, run this command and leave the window open. The first run downloads mradermacher’s 98.5 GB Qwen3.8-Flash-Next-Uncensored.IQ4_XS.gguf file; later runs reuse it. The server listens only on this computer at http://127.0.0.1:8080, names the model qwen3.8-flash-next-uncensored, and starts at 65,536 tokens of context with one request slot (-np 1), so the whole context goes to the agent. Hermes documents a 64K minimum context for local models, which is why this guide does not start lower. The sampling values are the ones OrcaRouter recommends for this model (temperature 1.0, top_p 0.95, top_k 20, min_p 0.0), and --jinja enables the Qwen tool-calling template. The --no-mmproj flag skips the vision file because this guide only works with text. The model thinks before answering by default. A download this large can take hours; if it is interrupted, run the same command again. Wait until the log says the server is listening.
Open a second PowerShell window on your own computer and run the official Hermes installer. Close and reopen PowerShell afterwards so the hermes command is available. This native installer does not need WSL.
Run these custom-provider settings in that PowerShell window. They are the same three settings Hugging Face shows in the Hermes Agent section of GGUF model pages, with this server’s address and the model name qwen3.8-flash-next-uncensored. Hermes then sends every request to llama.cpp on this computer.
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default qwen3.8-flash-next-uncensored
05
Create two small project files
In the same PowerShell window, create a disposable folder with a launch note and a checklist. The checklist deliberately has no database backup before the migration; the agent should find that gap from the files.
PowerShell
New-Item -ItemType Directory -Force "$HOME\agent-review" | Out-Null
Set-Location "$HOME\agent-review"
@"
Launch day: Thursday 14 May
Owner: Priya
Database migration runs at 09:00 before the app deploy.
"@ | Set-Content notes.md
@"
1. Freeze merges
2. Run the database migration
3. Deploy the app
4. Announce the release
"@ | Set-Content checklist.md
06
Give Hermes a file-based task
From inside the agent-review folder, start Hermes and paste the task below. Hermes asks before running commands or writing files; read each request before approving it. Approve the file reads and the single write of review.md. A model with reduced refusals will not push back on a risky action, so the approval prompt is your safeguard: keep it on.
Setting
Value
Task to paste into Hermes
Read notes.md and checklist.md in this folder. Write review.md containing: the launch day and owner, a corrected numbered checklist that adds a verified database backup before the migration, and one sentence explaining the risk of migrating without a backup. Cite the file each fact came from. Do not change notes.md or checklist.md.
Open review.md. It should name Thursday 14 May and Priya from notes.md, add a backup step before “Run the database migration”, and explain the risk. Check that Hermes’ activity shows real file reads and one write; an answer only in the chat window is not a completed agent run. Type /exit to leave Hermes and press Ctrl+C in the server window when you are done. To use the agent again, run the same server command (nothing is downloaded twice) and start hermes in any project folder.
03
SUCCESS CHECK · Qwen3.8 Flash Next Uncensored · IQ4_XS
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.
OrcaRouter’s card says tool calling works through the --jinja template and shows these Hermes custom-provider settings. The file used here is mradermacher’s quantization of that checkpoint, whose page gives no test results of its own. YouRunAI has not run this task with this file, so inspect the output rather than trusting the chat summary. Sources checked November 17, 2026.
Check that the server was started with --jinja, the model name is qwen3.8-flash-next-uncensored, and you started Hermes inside the agent-review folder. A chat-only answer is not a passing run.
The server stops with an out-of-memory error, or loading never finishes.
Close other apps that use the GPU or memory. On a PC whose GPU is smaller than the file, replace -ngl 99 with a lower number such as -ngl 20 so part of the model stays in system RAM; it will be slower. Lowering -c below 65536 saves memory, but Hermes may then warn that the context is too small for agent work.
llama-server says it does not recognise the model architecture.
This model uses the qwen4_exp architecture, which needs a llama.cpp build from August 27, 2026 or later. Update llama.cpp (winget upgrade llama.cpp, brew upgrade llama.cpp, or docker pull the server image again), check llama-server --version, and try again.
Answers take a long time to start, or stop before the final answer.
The model thinks before answering by default, and OrcaRouter advises allowing at least about 2,048 output tokens so the answer is not cut off. Thinking can be switched off per request with the enable_thinking chat-template setting.
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.