LLMBOTTLENECK.COM
Qwen / qwen3_5_moe_text

Qwen3.5-122B-A10B

125.1 billion parameters, routing 8 of 256 experts per token.

Open in the calculator →

Architecturepublished data
Quick answer
Qwen3.5-122B-A10B needs about 77.9 GB at Q4_K_M — the format most people download — with 8,192 tokens of context. It fits entirely on 5 of 18 common devices, starting with the AMD Ryzen AI Max+ 395 with Radeon 8060S (128 GB). What else fits in 128 GB. No hardware for it? Rent a GPU that runs it, priced live.

Sized at 8,192 tokens of context with the whole model in device memory, across 18 common devices. Speeds are estimates, not benchmarks.

Find the AMD Ryzen AI Max+ 395 with Radeon 8060S, the smallest card listed here that holds it: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

Weight-file size by format

FormatSizeRangeBasis
FP16250.3 GB175.1 GB – 252.7 GBsize range
Q8_0133.0 GB93.0 GB – 135.7 GBsize range
Q6_K102.7 GB71.8 GB – 105.5 GBsize range
Q5_K_M89.4 GB60.2 GB – 105.5 GBsize range
Q5_087.3 GB60.2 GB – 105.5 GBsize range
Q4_K_M76.8 GB49.3 GB – 105.5 GBsize range
Q4_072.8 GB49.3 GB – 105.5 GBsize range
Q3_K_M62.6 GB37.6 GB – 105.5 GBsize range
Q2_K49.5 GB28.7 GB – 105.5 GBsize range

A file size is not the memory a run needs: the KV cache and the runtime reserve come on top, and the calculator adds both.

What the labels mean
published data
Read from a published source — a file's byte count, a model's configuration or a manufacturer's specification — or exact arithmetic on such values. Not a measurement on a machine.
reconstructed size
Weight size reconstructed from the pinned architecture, because no published file exists.
size range
Only a lower and an upper bound are claimed for this weight size.

Hardware ladder

Q4_K_M at 8,192 tokens · 5 of 18 devices hold it entirely
DeviceMemoryVerdictDecodeCalculator
NVIDIA GeForce RTX 507077.9 GB of 12 GB42 layers on system RAM~14 tok/s with offloadOpen →
AMD Radeon™ RX 9070 XT77.9 GB of 16 GB39 layers on system RAM~15 tok/s with offloadOpen →
NVIDIA GeForce RTX 408077.9 GB of 16 GB39 layers on system RAM~15 tok/s with offloadOpen →
NVIDIA GeForce RTX 508077.9 GB of 16 GB39 layers on system RAM~15 tok/s with offloadOpen →
NVIDIA GeForce RTX 5070 Ti77.9 GB of 16 GB39 layers on system RAM~15 tok/s with offloadOpen →
NVIDIA GeForce RTX 5060 Ti 16GB77.9 GB of 16 GB39 layers on system RAM~15 tok/s with offloadOpen →
NVIDIA GeForce RTX 409077.9 GB of 24 GB34 layers on system RAM~17 tok/s with offloadOpen →
AMD Radeon™ RX 7900 XTX77.9 GB of 24 GB34 layers on system RAM~17 tok/s with offloadOpen →
NVIDIA GeForce RTX 309077.9 GB of 24 GB34 layers on system RAM~17 tok/s with offloadOpen →
NVIDIA GeForce RTX 509077.9 GB of 32 GB29 layers on system RAM~20 tok/s with offloadOpen →
NVIDIA RTX 6000 Ada Generation77.9 GB of 48 GB19 layers on system RAM~28 tok/s with offloadOpen →
Apple M4 Pro77.9 GB of 64 GBdoes not fitnot calibratedOpen →
Apple M5 Pro77.9 GB of 64 GBdoes not fitnot calibratedOpen →
AMD Ryzen AI Max+ 395 with Radeon 8060S77.9 GB of 128 GBfits~39 tok/sOpen →
Apple M3 Max77.9 GB of 128 GBfits~52 tok/sOpen →
Apple M4 Max77.9 GB of 128 GBfits~64 tok/sOpen →
NVIDIA DGX Spark77.9 GB of 128 GBfits~37 tok/sOpen →
Apple M2 Ultra77.9 GB of 192 GBfits~80 tok/sOpen →

Decode is a calibrated estimate from the published calibration; capacity is the manufacturer's published ceiling, not guaranteed free memory. Fitting in memory is not the same as loading: whether the runtime and version you have supports this architecture and format on that machine has not been tested here. Offloaded rows assume there is enough system RAM for the overflow — the calculator checks that against the RAM you declare. Each “Open” link keeps this model, format and context.

Run it in the cloud

Rent a GPU that runs Qwen3.5-122B-A10B

Every rentable machine that holds the whole model, from Vast.ai and RunPod, with this site's speed estimate for each and the provider's own price, read live. Cheapest first; sort by cost per token or speed instead, or switch the format.

Prices · 18:30 UTC, 22 Sept
18 configurations in stock · 13 out of stock
MachineSpeedPer hourPer M tokensWhereRent
8× RTX 306096 GBcheapest~289 tok/sFaster than you read$0.59$0.57Vast.aiMarketplace · 99.7% reliableRent on Vast.ai
4× RTX 309096 GB~443 tok/sFaster than you read$0.60$0.37Vast.aiMarketplace · 99.4% reliableRent on Vast.ai
4× RTX 5090128 GBrecommendedbest value~849 tok/sFaster than you read$1.07$0.35Vast.aiMarketplace · 99.6% reliableRent on Vast.ai
2× RTX 6000 Ada96 GB~250 tok/sFaster than you read$1.07$1.19Vast.aiMarketplace · 99.4% reliableRent on Vast.ai
1× RTX PRO 600096 GB~240 tok/sFaster than you read$1.14$1.31Vast.aiMarketplace · 98.8% reliableRunPod $1.69/h Rent on Vast.ai
1× A100 80GB PCIe80 GB~259 tok/sFaster than you read$1.19$1.27RunPodCommunity CloudRent on RunPod
2× RTX A600096 GB~200 tok/sFaster than you read$1.20$1.67Vast.aiMarketplace · 99.4% reliableRent on Vast.ai
4× L496 GB~142 tok/sFaster than you read$1.29$2.51Vast.aiMarketplace · 99.4% reliableRent on Vast.ai

No GPU at all

Use Qwen3.5-122B-A10B by the token

Here the rented card is as cheap as the API or cheaper per token ($0.35 per million), and it keeps your data on a machine you control. The API is still the easier start: nothing to set up and nothing to switch off.

First time renting a GPU? How it works, in four steps
  1. Create an account and add credit. Both providers are prepaid: $10 is enough to try any card on this page for hours. A new RunPod account that arrives through a referral link gets a one-time $5 credit when it first adds $10.
  2. Pick the card and the count shown here, and a template: llama.cpp or vLLM for an OpenAI-compatible API, or one with a web chat if you only want to talk to the model. “How to launch it” below gives the exact image and command.
  3. Wait for the download. The weights are fetched on the machine; tens of gigabytes take a few minutes on a data-centre connection. You pay by the second from the moment it starts.
  4. Stop it when you are done. A running machine bills even when idle. A stopped one costs nothing per hour, though its disk is usually still billed until you delete it.
How to launch it

No Q4_K_M file of Qwen3.5-122B-A10B is catalogued here, so there is no exact command to vouch for. Start the machine from the llama.cpp server image ghcr.io/ggml-org/llama.cpp:server-cuda and point -hf at a Q4_K_M build from the publisher or a quantizer you trust — search Hugging Face for one. Check its file size against the memory figure on this page before you rent.

Every machine holds the whole model at Q4_K_M and 8,192 tokens of context, no offload. Speeds are this site’s single-stream decode estimates; prices are what each provider’s own API quoted, on-demand, for the whole machine. Vast hosts below 98% measured reliability are left out.

Referral links Vast.ai, RunPod and Novita pay us a share of what you spend if you sign up through these buttons. It costs you nothing, and it never decides an order or a recommendation: both are computed from the live price and the speed, and options that pay us nothing are listed and recommended on the same terms. How we rank

Buying a card instead, or paying by the token? Compare a year of Qwen3.5-122B-A10B three ways →

Under the hood

Architecture, read from the publisher’s file

The numbers every figure above is computed from, with the file they came from.

Architecture

✓ Architecture read from the published config.json

Retrieved 2026-09-01 at pinned commit dc4d348443bc; parameter count from the safetensors index at the same revision.

Architecture
qwen3_5_moe_text
Layers
48
Hidden size
3,072
Attention heads
32
KV heads
2
Head dimension
256
Vocabulary
248,320
Context ceiling
262,144
RoPE theta
10,000,000
Experts
256
Experts / token
8
Expert width
1,024
Shared expert width
1,024

Where the memory goes

tokenembedding× 48 decoder blocksGrouped-query attention32 query · 2 KV headsfull contextRouted experts8 of 256 per tokenall residentoutputprojectiongrows with contextcapacity ≠ traffic

Grouped-query attention shares each key/value head across 16 query heads, so the KV cache is 6% of what multi-head attention would need at the same context.

This is a multimodal checkpoint (vision); the diagram and memory sizing above cover the text decoder. Encoder/audio/image activations are not included in the KV or weight figures.

Open the exact official config.json ↗

More from Qwen

Qwen-AgentWorld-35B-A3BQwen2.5-0.5BQwen2.5-0.5B-InstructQwen2.5-1.5B-InstructQwen2.5-14B-InstructQwen2.5-32B-InstructQwen2.5-3B-InstructQwen2.5-72B-Instruct

Use this answer in your own product

As JSON, from the API
curl -s https://llmbottleneck.com/v1/analyze \
  -H "Authorization: Bearer $LLMB_KEY" -H "content-type: application/json" \
  -d '{"model":"qwen-qwen3-5-122b-a10b","quantization":"Q4_K_M","context":8192,"hardware":"nvidia-geforce-rtx-4090"}'

Same engine, same evidence, every field sourced. Free key in one step →

On your page, as a widget
<script src="https://llmbottleneck.com/widget.js"
  data-model="qwen-qwen3-5-122b-a10b" data-quantization="Q4_K_M"
  data-hardware="nvidia-geforce-rtx-4090" data-context="8192"></script>

No key needed. Unbranded, with your own buy button, on Pro and Business →

Citing this page

LLM Bottleneck. “Qwen3.5-122B-A10B VRAM and hardware requirements.” Architecture from Qwen/Qwen3.5-122B-A10B at revision dc4d348443bc, retrieved 2026-09-01. https://llmbottleneck.com/models/qwen-qwen3-5-122b-a10b

Every figure above is either the published value or a reconstruction whose measured error is on the accuracy page.

llmbottleneck
catalogue 2026-10-03models 327devices 135