Skip to content
Llama Builds

Alibaba Qwen · Released February 2026

Qwen3.5 122B-A10B

What it takes to run Qwen3.5 122B-A10B locally: memory by quantization, the GPUs that handle it, and how fast it runs.

Updated

Parameters
122B (10B active)
Architecture
Mixture of experts
Context
256K tokens
License
Apache 2.0 license
Memory at Q4_K_M
~77GB

What do you need to run Qwen3.5 122B-A10B?

Qwen3.5 122B-A10B needs about 77GB of VRAM at Q4_K_M with 8K context (74GB for the weights plus KV cache and overhead). The smallest GPU that runs it comfortably at Q3_K_M is a 72GB card like the RTX PRO 5000 72GB. Because it is a mixture-of-experts model with 10B active parameters, it generates text much faster than a dense model of the same size, and it tolerates partial CPU offload better.

See the full hardware guide for Qwen3.5 122B-A10B

What to buy

GPUs for Qwen3.5 122B-A10B

Unified memory

Ryzen AI Max+ 395

128GB unified · ~$3,300

Quant
Q4_K_M
Speed (est.)
~18 tok/s
Runs well
Check price at Amazon

Multi-GPU

4× RTX 3090

96GB total · ~$4,200 used

Quant
Q4_K_M
Speed (est.)
~39 tok/s
Runs well
Check price at Amazon

Single card

RTX PRO 5000 72GB

72GB · ~$9,100

Quant
Q3_K_M
Speed (est.)
~99 tok/s
Runs well
Check price at Amazon

Speeds are estimates at the quantization shown with 8K context. We may earn a commission when you buy through these links. Learn more.

Memory

Qwen3.5 122B-A10B VRAM requirements by quantization

Lower-bit quantizations shrink the model at some cost to quality. Q4_K_M is the usual sweet spot. Totals include the KV cache for the context length shown plus runtime overhead.

Memory needed for Qwen3.5 122B-A10B at each quantization level
QuantizationQualityWeightsTotal @ 8K ctxTotal @ 32K ctx
Q2_KRough46 GB48 GB49 GB
Q3_K_MUsable59 GB62 GB63 GB
Q4_K_MRecommendedDefault74 GB77 GB78 GB
Q5_K_MVery good87 GB90 GB91 GB
Q6_KNear-lossless101 GB104 GB105 GB
Q8_0Practically lossless130 GB134 GB135 GB
BF16Full precision244 GB252 GB253 GB

KV cache note: Hybrid layout: three of every four layers use Gated DeltaNet linear attention with a fixed-size state. Assuming 48 layers with 12 full-attention layers of 2 KV heads x 256 dims: 2 x 12 x 2 x 256 x 2 bytes = 24,576 bytes per token, about 6.4GB at the full 262K.

Performance

How fast Qwen3.5 122B-A10B runs

Single-user generation speed on popular GPUs that fit the model entirely in memory, at the best quantization each can hold.

  1. RTX PRO 6000

    Q4_K_M · 96GB

    ~105tok/s

  2. RTX PRO 5000 72GB

    Q3_K_M · 72GB

    ~99tok/s

  3. M5 Max 128GB

    Q4_K_M · 128GB

    ~34tok/s

  4. M5 Ultra 512GB

    Q8_0 · 512GB

    ~33tok/s

  5. M5 Ultra 256GB

    Q8_0 · 256GB

    ~33tok/s

  6. M4 Max 128GB

    Q4_K_M · 128GB

    ~31tok/s

  7. M3 Ultra 512GB

    Q8_0 · 512GB

    ~23tok/s

  8. Ryzen AI Max+ 395

    Q4_K_M · 128GB

    ~18tok/s

  9. DGX Spark

    Q6_K · 128GB

    ~14tok/s

  10. Ryzen AI Max+ PRO 495

    Q6_K · 192GB

    ~14tok/s

Estimated (method)

Overview

About Qwen3.5 122B-A10B

Qwen3.5 122B-A10B is the mid-size mixture-of-experts model from Qwen's February 2026 Qwen3.5 launch, with 10B active parameters, image input, a 262K context, and switchable thinking. Later Qwen releases refreshed the 27B and 35B-A3B sizes, so this remains Qwen's main model for 96GB to 128GB machines. It needs about 77GB at Q4_K_M with 8K context, which suits Macs with 96GB or more, Strix Halo 128GB systems, and DGX Spark; a 24GB GPU needs 96GB or more of system RAM to hold the offloaded experts. Its hybrid attention keeps the KV cache small, so long contexts are cheap.

Strengths

  • +Strong all-round model with image input and toggleable thinking
  • +10B active parameters keep it fast on unified-memory machines
  • +Small KV cache thanks to hybrid linear attention
  • +Apache 2.0 license

Weaknesses

  • −Needs about 77GB at Q4_K_M, beyond any single consumer GPU
  • −Smaller 2026 models such as Qwen3.8 27B now score about as well on benchmarks
  • −Hybrid attention needs recent runtime versions
ChatReasoningCodingAgents & toolsVisionLong contextMultilingualImage inputToggleable reasoning

Setup

How to run Qwen3.5 122B-A10B

Ollama
$ ollama run qwen3.5:122b
llama.cpp
$ llama-server -hf unsloth/Qwen3.5-122B-A10B-GGUF:Q4_K_M -c 8192 -ngl 99

Weights: Qwen/Qwen3.5-122B-A10B · GGUF: unsloth/Qwen3.5-122B-A10B-GGUF

Alternatives

Models like Qwen3.5 122B-A10B

OpenAI

gpt-oss-120b

117B (5.1B active) · 128K ctx

~68GBat MXFP4
ReasoningCodingAgents & tools

Z.ai

GLM-4.5-Air

106B (12B active) · 128K ctx

~68GBat Q4_K_M
CodingAgents & toolsReasoning

Alibaba Qwen

Qwen3.8 27B

27B · 256K ctx

~18GBat Q4_K_M
CodingAgents & toolsReasoning

DeepSeek

DeepSeek V4 Flash

284B (13B active) · 1.0M ctx

~168GBat FP4
ReasoningCodingAgents & tools

Frequently asked questions

How much VRAM do you need to run Qwen3.5 122B-A10B?

About 77GB at Q4_K_M with 8K tokens of context: 74.0GB of weights, 0.2GB of KV cache, and 2.8GB of runtime overhead. Higher-precision quants and longer contexts need more; see the table above.

What is the cheapest GPU that can run Qwen3.5 122B-A10B?

The smallest card in our database that runs it comfortably is the RTX PRO 5000 72GB (72GB), at Q3_K_M and ~99 tokens/sec.

Is Qwen3.5 122B-A10B free for commercial use?

Qwen3.5 122B-A10B is released under the Apache 2.0 license. That is a permissive license, so commercial use is allowed.

How long a context does Qwen3.5 122B-A10B support?

Up to 262,144 tokens (256K). Long contexts need a lot of extra memory for the KV cache, so most local setups run 8K to 32K.

Sources: Artificial Analysis: Qwen3.5 122B A10B, Unsloth Qwen3.5 122B-A10B GGUF files, modelfit.io Qwen3.5 122B-A10B VRAM requirements. Capability score basis: Estimated level with Qwen3.5 27B (42) on the Artificial Analysis Intelligence Index v4 (February to April 2026), with a small allowance for broader knowledge; no exact reading was confirmed.