GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
MarMix
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
MarMix
Model Tree
LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Final-GGUF
Input Modalities
Output Modalities
Supported Functionality
Genesis release:
https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Final-GGUF
Genesis is based on:
https://huggingface.co/HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive
Reference architecture files / missing architecture tensors used during reconstruction:
https://huggingface.co/Qwen/Qwen3.6-35B-A3B
Please see the upstream Genesis repository for the author's description of the Genesis process, acknowledgements, intended behavior, recommended prompting, and original GGUF releases.
The conversion was performed with the Qwen3.6 MoE conversion path and then verified against both the source GGUF and the Hugging Face reference layout.
Verification summary:
The generated checkpoint contains 17 SafeTensors shards.
The reconstructed model resolves in current Transformers/vLLM as:
Qwen3_5MoeForConditionalGeneration
Key upstream architecture characteristics:
The included Hugging Face processor loads as:
Qwen3VLProcessor
with image and video processors available.
The upstream GGUF release uses a separate mmproj file for llama.cpp-style multimodal inference.
Do not load the GGUF mmproj file with this SafeTensors/vLLM reconstruction.
For this repository, the Hugging Face multimodal processor and the required vision tensors are already part of the reconstructed Transformers-compatible model layout.
The vision/MTP tensors that were absent from the GGUF representation came from the compatible Qwen/Qwen3.6-35B-A3B reference model. They should not be represented as Genesis-modified tensors unless independently demonstrated.
Validated environment:
The following configuration is recommended for interactive single-user use with OpenCode:
bash
# Use 127.0.0.1 for local-only access.# Use 0.0.0.0 only if you intentionally want vLLM reachable from other machines.vllm serve /path/to/Qwen3.6-35B-A3B-Uncensored-Genesis-Final-Safetensors-BF16 \ --host 127.0.0.1 \ --port 11620 \ --served-model-name Qwen3.6-35B-A3B-Uncensored-Genesis-Final \ --dtype bfloat16 \ --max-model-len 131072 \ --gpu-memory-utilization 0.80 \ --max-num-seqs 32 \ --reasoning-parser qwen3 \ --enable-auto-tool-choice \ --tool-call-parser qwen3_xml--host 127.0.0.1 keeps the API accessible only from the local machine.--host 0.0.0.0 makes vLLM listen on all network interfaces. Use this only when remote access is required, and protect the server with an API key and appropriate firewall rules.--max-model-len 131072 enables a maximum context length of 128K tokens.--gpu-memory-utilization 0.80 leaves additional VRAM headroom for CUDA workspaces, temporary allocations, multimodal processing, and long-running workloads. The model was previously validated successfully with 0.90, but 0.80 is a more conservative workstation setting.--max-num-seqs 32 is recommended for single-user OpenCode use. The model was previously tested with 512, which is more appropriate for high-concurrency server workloads.--reasoning-parser qwen3 enables Qwen-style reasoning parsing.For a local OpenAI-compatible endpoint that should not be anonymously accessible, also add:
bash
--api-key YOUR_API_KEYBinding to 127.0.0.1 is still recommended for local-only use.
On the NVIDIA RTX PRO 6000 Blackwell Workstation Edition, vLLM may perform first-run compilation, CUDA graph capture, and FlashInfer/CUTLASS autotuning.
That initial startup can take substantially longer than later starts because compiled kernels and autotuning results are cached.
During validated startup:
High VRAM allocation while idle is therefore expected with --gpu-memory-utilization 0.90; it does not mean the GPU is under heavy compute load.
With the validated 131072-token configuration, vLLM reported approximately:
Actual throughput and concurrency depend on prompt length, output length, multimodal inputs, sampling, and concurrent requests.
The upstream Genesis README recommends the following for thinking-mode coding / precise tasks:
text
temperature = 0.6top_p = 0.95top_k = 20min_p = 0.0seed = 42presence_penalty = disabledrepeat_penalty = disabledThe following settings are recommended by the upstream Genesis release.
Thinking mode (coding):
temperature=0.6, top_p=0.95, top_k=20, min_p=0.0, seed=42, presence_penalty=disabled, repeat_penalty=disabledtemperature=0.95, top_p=0.95, top_k=20, min_p=0.0, seed=42, presence_penalty=disabled, repeat_penalty=disabledtext
temperature = 0.6top_p = 0.95top_k = 20min_p = 0.0seed = 42presence_penalty = disabledrepeat_penalty = disabledNon-thinking mode (creative):
temperature=0.7, top_p=0.8, top_k=20, min_p=0.0, seed=42, presence_penalty=disabled, repeat_penalty=disabledtemperature=0.95, top_p=disabled, top_k=20, min_p=disabled, seed=42, presence_penalty=disabled, repeat_penalty=disabledtext
temperature = 0.7top_p = 0.8top_k = 20min_p = 0.0seed = 42presence_penalty = disabledrepeat_penalty = disabledA neutral repeat penalty of 1.0 is equivalent to applying no repetition penalty in runtimes that require a numeric value.
The upstream recommended minimal system prompt is:
text
You are Qwen, a large language model developed by Alibaba Group's Tongyi Lab. You are a helpful assistant.The upstream author also recommends keeping at least 128K context for thinking behavior.
Example provider configuration for an OpenAI-compatible vLLM endpoint:
json
{ "provider": { "vllm-genesis": { "npm": "@ai-sdk/openai-compatible", "name": "Qwen3.6 Genesis Final BF16 via vLLM", "options": { "baseURL": "http://127.0.0.1:10101/v1", "apiKey": "{file:/home/openrag/.config/opencode/vllm-genesis-api-key}" }, "models": { "Qwen3.6-35B-A3B-Uncensored-Genesis-Final": { "name": "Qwen3.6-35B-A3B Uncensored Genesis Final BF16", "limit": { "context": 131072, "output": 32768 }, "tool_call": true, "temperature": true, "reasoning": true, "attachment": true } } } }, "model": "vllm-genesis/Qwen3.6-35B-A3B-Uncensored-Genesis-Final"}Recommended OpenCode coding-agent sampling profile Thinking mode (coding):
json
{ "temperature": 0.6, "top_p": 0.95, "top_k": 20, "min_p": 0.0, "presence_penalty": 0.0, "repetition_penalty": 1.0, "seed": 42}This repository is:
This repository is not:
mmproj.gguf when used through Transformers/vLLMAll model-development, fine-tuning/uncensoring, Genesis processing, and upstream release credit belongs to the respective upstream authors.
This repository only provides the SafeTensors mixed-precision reconstruction and verification described above.
Please refer to the upstream repositories for their full credits, licensing terms, model behavior notes, and recommended usage.
--enable-auto-tool-choice allows the model to decide when to call tools.--tool-call-parser qwen3_xml enables parsing of Qwen tool calls for OpenAI-compatible clients such as OpenCode.--gpu-memory-utilization and --max-num-seqs may be appropriate. For an interactive workstation, lower values provide more VRAM headroom and reduce unnecessary resource pressure.