AtomicChat/Qwen3.6-35B-A3B-GGUF

Base model: Qwen/Qwen3.6-35B-A3B

Qwen3.6 35B A3B, self-quantized to GGUF by Atomic Chat. Built straight from Qwen's original weights with a per-tensor importance matrix, so this is not a repack of somebody else's files. Runs fully offline.

Highlights

36.0B parameters: the weights this repo quantizes.
Context length: 262,144 tokens (256K), as published by Qwen.
40 layers: Mixture-of-Experts.
Modalities: the base model handles Text, Image; this repo ships text-only quants, it carries no vision projector.
Full imatrix ladder: every quant is calibrated with an importance matrix.
Agentic Coding:: the model now handles frontend workflows and repository-level reasoning with greater fluency and precision.
Thinking Preservation:: we've introduced a new option to retain reasoning context from historical messages, streamlining iterative development and reducing overhead.

[!NOTE] These GGUFs are self-quantized from the original weights, not a repack. The importance matrix keeps low-bit quants closer to the full-precision model.

[!IMPORTANT] Always pass --jinja so the Qwen3.6 35B A3B chat template is applied. Without it the model can emit malformed turns.

Model Overview

Property	Value
Base model	`Qwen/Qwen3.6-35B-A3B`
Parameters	36.0B
Layers	40
Experts	256 routed (top-8)
Context length	262,144 tokens (256K)
Vocabulary	248,320
Modalities	Text, Image in the base model; text only in this repo, it ships no vision projector
Architecture	Mixture-of-Experts, 256 experts (top-8), 16 attention heads over 2 KV heads, `Qwen3_5MoeForConditionalGeneration`
This repo	GGUF quants (imatrix). Quants: `Q4_K_M`, `UD-Q4_K_XL`, `Q5_K_M`, `Q6_K`, `Q8_0`

Scores are Qwen's published results for the base Qwen/Qwen3.6-35B-A3B, not our own measurements. Quantization preserves the large majority of this; Q4_K_M and up stay close to full precision.

Choosing a quant

Quant	Size	Notes
`Q4_K_M`	21.2 GB	Recommended default. Best balance of size, speed and quality.
`UD-Q4_K_XL`	21.5 GB	Dynamic. Embeddings and output kept at Q8_0 for higher quality at a Q4 footprint.
`Q5_K_M`	24.7 GB	Higher quality, low loss.
`Q6_K`	28.5 GB	Near lossless, noticeably lighter than Q8_0.
`Q8_0`	36.9 GB	Effectively lossless, reference quality.

[!TIP] Pick the largest file that fits your (V)RAM with room for context. Q4_K_M or UD-Q4_K_XL is the sweet spot for most setups; Q6_K or Q8_0 for maximum fidelity.

Get started

Run Qwen3.6 35B A3B locally with:

Atomic Chat: the easiest path. Open the app, search AtomicChat/Qwen3.6-35B-A3B-GGUF, pick a quant, hit Use this model.
llama.cpp: llama-server -hf AtomicChat/Qwen3.6-35B-A3B-GGUF:Q4_K_M --jinja -c 8192
Ollama: ollama run hf.co/AtomicChat/Qwen3.6-35B-A3B-GGUF:Q4_K_M
LM Studio / Jan: search the repo id, download any quant.

Best practices

Parameter	Value
temperature	1.0
top_p	0.95
top_k	20
min_p	0.0
repetition_penalty	1.0

Qwen's recommended sampling configuration for Qwen/Qwen3.6-35B-A3B.

Run in llama.cpp

git clone https://github.com/ggml-org/llama.cpp
cmake llama.cpp -B llama.cpp/build -DBUILD_SHARED_LIBS=OFF -DGGML_CUDA=ON
cmake --build llama.cpp/build --config Release -j --target llama-cli llama-server

./llama.cpp/build/bin/llama-server \
    -hf AtomicChat/Qwen3.6-35B-A3B-GGUF:Q4_K_M \
    --jinja -ngl 99 -c 8192 -fa on

How these were made

Download Qwen/Qwen3.6-35B-A3B (original weights).
Convert to f16 GGUF with llama.cpp.
Build an importance matrix over our calibration corpus.
Quantize the ladder with --imatrix.
UD-Q4_K_XL additionally pins the token-embedding and output tensors to Q8_0.

License

Original model by Qwen, released under the Apache 2.0 license. Full terms: Apache 2.0. Quantized by Atomic Chat.