llmrun
codacusQwen 3.6

Qwen3.6 35B A3B MicroTurn Duplex LoRA — Hardware Requirements & GPU Compatibility

Chat

Qwen3.6 35B A3B MicroTurn Duplex LoRA is a 35B-parameter open language model from codacus in the Qwen 3.6 family. At Q4_K_M it needs about 23.10 GB of VRAM — see which GPUs and Macs can run it below.

35B params · MoE3B activeApache 2.0133 downloads27 likes
Based on Qwen3.6 35B A3B
B

Decent on a 24 GB GPU

Runs on24 GB Q4_K_M · ~177 tok/s · 4KMacs from32 GB ~30 tok/s

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Qwen3.6 35B A3B MicroTurn Duplex LoRA

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Recommended

vLLM
one command · needs vLLM installed
$vllm serve codacus/Qwen3.6-35B-A3B-MicroTurn-Duplex-LoRA --max-model-len 16384

vLLM repo with its own quantisation

Download: Weights on Hugging Face

install, OS, other quants

Other ways

each opens step-by-step

Run in cloud

Fits on RTX A6000 (48 GB) (21 GB headroom) · Q4_K_M

Generation speed
~151 tok/s
generation speed
Cost per 1M output tokens
$0.74
per 1M output tokens
Deployon Vast.aiBest optionfrom $0.40/hr (opens in new tab)
Compare GPUs →
or

How Much VRAM Does Qwen3.6 35B A3B MicroTurn Duplex LoRA Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KOn RTX 4090 (reference)
16.4 GB24.7 GB2-bit · K-quant — 2-bit quantization with K-quant improvementsFits: 16K context · ~209 tok/s
18.8 GB27.1 GB3-bit medium — 3-bit medium quantizationFits: 16K context · ~196 tok/s
23.1 GB31.4 GB4-bit medium — 4-bit medium quantization — most popular sweet spotFits: 4K context · ~177 tok/s
27.4 GB35.7 GB5-bit medium — 5-bit medium quantization — good quality/size tradeoffDoes not fit: needs 31.3 GB
31.8 GB40.1 GB6-bit, very good — 6-bit quantization, very good qualityDoes not fit: needs 35.6 GB
38.5 GB46.8 GB8-bit, near-lossless — 8-bit quantization, near-losslessDoes not fit: needs 42.4 GB
77 GB85.3 GB16-bit · bfloat — Brain floating point 16 — preferred for trainingDoes not fit: needs 80.9 GB
est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

27.0GBat 16K context

Weights
22.3 GB
KV cache
4.4 GB
Overhead
0.3 GB

Fits RTX 4090 (reference) · fits up to 4K context · ~177 tok/s

≥50 tok/s needs ≥175 GB/s · RTX 5090 ~230, M3 Ultra ~77

Or rent RTX A6000 from $0.40/hr.

Which Hardware Can Run Qwen3.6 35B A3B MicroTurn Duplex LoRA?

Q4_K_M · 23.1 GB

Qwen3.6 35B A3B MicroTurn Duplex LoRA (Q4_K_M) requires 23.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 31+ GB is recommended. 7 GPUs can run it, including NVIDIA GeForce RTX 5090. 103 devices with unified memory can run Qwen3.6 35B A3B MicroTurn Duplex LoRA, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Studio M5 Max (36 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M
8–16 GB · RTX 4060, RTX 3060 12GB, RTX 5060 Ti 16GBDoesn't fit — smallest quant needs 20.2 GB8 GB tier →12 GB tier →16 GB tier →
24 GB · RTX 4090referenceQ4_K_M · 27 GB · ~177 tok/s · 4Ktight24 GB tier →
Mac 32 GB · Mac Mini M4Q4_K_M · 27 GB · ~30 tok/s · 16KMacs →
Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 27 GB · ~68 tok/s · 32KMacs →

Alternatives to Qwen3.6 35B A3B MicroTurn Duplex LoRA

Qwen 3.6 family · 15 models →

Similar models

6
ModelScoreVRAMtok/sWhy it is similar
Llm Jp 4.1 33B Thinkingllm-jp—22.5 GB~35same size
Agnes 3.0 FlashAgnes-AI—22.1 GB~35same size
Ornith 1.5 35B A3Bornith-ai—22 GB~151same size
Agents A1InternScience—21.5 GB~171same size
Intern S2 PreviewInternLM—73.2 GB—needs > 24 GBsame size
EXAONE 4.5 33BLGAI-EXAONE—23.9 GB~33same size

Specifications

Publisher
codacus
Family
Qwen 3.6
Parameters
35B
Active parameters
3B per token
Release Date
2026-10-12
License
Apache 2.0

Resources

Frequently Asked Questions

How much VRAM does Qwen3.6 35B A3B MicroTurn Duplex LoRA need?

Qwen3.6 35B A3B MicroTurn Duplex LoRA requires 23.1 GB of VRAM at Q4_K_M, or 77 GB at BF16.

VRAM = Weights + KV Cache + Overhead

Weights = 35B × 4.8 bits ÷ 8 = 21 GB

KV Cache + Overhead ≈ 2.1 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

VRAM usage by quantization

23.1 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 4090 run Qwen3.6 35B A3B MicroTurn Duplex LoRA?

Yes, at Q4_K_M (23.1 GB) or lower. Higher quantizations like Q5_K_M (27.4 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.

What's the best quantization for Qwen3.6 35B A3B MicroTurn Duplex LoRA?

For Qwen3.6 35B A3B MicroTurn Duplex LoRA, Q4_K_M (23.1 GB) offers the best balance of quality and VRAM usage. Q5_K_M (27.4 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 16.4 GB.

VRAM requirement by quantization

Q2_K
16.4 GB
Q4_K_M ★
23.1 GB
Q5_K_M
27.4 GB
Q6_K
31.8 GB
Q8_0
38.5 GB
BF16
77.0 GB

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Qwen3.6 35B A3B MicroTurn Duplex LoRA on a Mac?

Yes, but only at lower quantizations. The smallest Mac that can run Qwen3.6 35B A3B MicroTurn Duplex LoRA is Mac Mini M2 (24 GB) at Q2_K; 107 of the 156 Macs we list can run it at some quantization. For Q4_K_M (23.1 GB) you need a Mac with more unified memory.

Can I run Qwen3.6 35B A3B MicroTurn Duplex LoRA locally?

Yes — Qwen3.6 35B A3B MicroTurn Duplex LoRA can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Qwen3.6 35B A3B MicroTurn Duplex LoRA?

At Q4_K_M, Qwen3.6 35B A3B MicroTurn Duplex LoRA can reach ~100 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~177 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)

Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.

Example: NVIDIA B200 → 2.0 GB active ÷ (8000 × 0.65) = 0.38 ms, plus 48 layers × 0.055 ms = 2.64 ms, so 1000 ÷ 3.02 ms = ~331 tok/s

Estimated speed at Q4_K_M (23.1 GB)

~331 tok/s
~177 tok/s
~331 tok/s
~307 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Qwen3.6 35B A3B MicroTurn Duplex LoRA?

At Q4_K_M, the download is about 21.00 GB. The full-precision BF16 version is 70.00 GB. The smallest option (Q2_K) is 14.88 GB.

Which GPUs can run Qwen3.6 35B A3B MicroTurn Duplex LoRA?

7 consumer GPUs can run Qwen3.6 35B A3B MicroTurn Duplex LoRA at Q4_K_M. Top options include AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090.

Which devices can run Qwen3.6 35B A3B MicroTurn Duplex LoRA?

103 devices with unified memory can run Qwen3.6 35B A3B MicroTurn Duplex LoRA at Q4_K_M (23.1 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB), Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.