Text model · Qwen3.6
Qwen-AgentWorld 35B-A3B: RAM and VRAM requirements
Qwen-AgentWorld 35B-A3B needs about 24.3 GB to run at Q4_K_M (the Q4_K_M GGUF file is ~22.13 GB to download; KV cache and overhead add the rest), or about 39.1 GB at Q8_0. The lightest hardware that runs it is Nvidia GeForce RTX 5090 (32GB).
Qwen3.6 family · 34.7B params (Mixture-of-Experts: activates only 3B of 34.7B params per token, so generation is faster than the total size suggests) · released Jun 2026.
Shopping for hardware? See what runs Qwen-AgentWorld 35B-A3B →
Q4_K_M and Q8_0 are GGUF download (file) sizes. Memory @ Q4 includes the KV cache and overhead, so it runs larger than the file on disk.
Will it run on your device?
Qwen-AgentWorld 35B-A3B runs on 9 of 40 tracked devices at Q4_K_M.
Memory breakdown
How context length changes it
Longer context grows the KV cache, which for Qwen-AgentWorld 35B-A3B is sized by its 3B active params, not the full 34.7B. It needs ~24.3 GB at 4k and ~68.1 GB at 128k.
Quantization sizes
| Quantization | Size on disk |
|---|---|
| Q2_K | 14.5 GB est |
| Q3_K_M | 17 GB est |
| Q4_K_M (default) | 22.13 GB |
| Q5_K_M | 24.7 GB est |
| Q6_K | 28.5 GB est |
| Q8_0 | 36.9 GB |
| FP16 | 69.38 GB |
Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.
Run it
llama-cli -hf unsloth/Qwen-AgentWorld-35B-A3B-GGUF:Q4_K_M lms get unsloth/Qwen-AgentWorld-35B-A3B-GGUF Which devices can run Qwen-AgentWorld 35B-A3B?
Apple Silicon Macs
- Apple M1 (8GB) No
- Apple M2 (16GB) No
- Apple M4 (16GB) No
- Apple M5 (16GB) No
- Apple M3 Pro (18GB) No
- Apple M4 (24GB) No
- Apple M4 Pro (24GB) No
- Apple M5 (32GB) No
- Apple M4 Pro (48GB) Yes
- Apple M5 Pro (48GB) Yes
- Apple M4 Max (64GB) Yes
- Apple M4 Max (128GB) Yes
- Apple M5 Max (128GB) Yes
- Apple M3 Ultra (256GB) Yes
RAM-only laptops
iPhone & iPad
Android
NVIDIA GPUs
Same job, different size
Pick by what fits your memory: step down to free up VRAM, or step up if you have headroom.
Similar models
Head-to-head
FAQ
How much VRAM or RAM does Qwen-AgentWorld 35B-A3B need?
At Q4_K_M, Qwen-AgentWorld 35B-A3B needs about 24.3 GB (weights ~22.13 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~39.1 GB.
What is the Q4_K_M GGUF file size of Qwen-AgentWorld 35B-A3B?
The Q4_K_M GGUF file is about 22.13 GB to download, and the Q8_0 GGUF is about 36.9 GB. That is the weights file on disk; to run it you also need room for the KV cache and overhead, so budget ~24.3 GB of memory at Q4_K_M.
Can Qwen-AgentWorld 35B-A3B run on a laptop?
Qwen-AgentWorld 35B-A3B is large; you need a high-memory Mac or multi-GPU setup at Q4_K_M.
Is Qwen-AgentWorld 35B-A3B cheaper to run because it is a MoE model?
It is faster, not lighter. Qwen-AgentWorld 35B-A3B activates only 3B of 34.7B params per token (so it runs quickly), but all experts must stay in memory, so it still needs memory for the full 34.7B.
Can I use Qwen-AgentWorld 35B-A3B commercially?
Yes. Qwen-AgentWorld 35B-A3B is licensed Apache-2.0, which permits commercial use.
Understand the numbers
Short guides to the ideas behind Qwen-AgentWorld 35B-A3B's memory and quant figures.
Alibaba's agent 'world model' (June 2026) built on the Qwen3.6 35B-A3B MoE base: trained on 10M+ real environment trajectories across MCP, search, terminal, SWE, web, OS and Android domains. Vision-capable (mmproj required). Q4_K_M is unsloth's UD-Q4_K_M dynamic quant; Q8_0/BF16 are plain quants from the same repo. Not on Ollama's library; run via llama.cpp or LM Studio.
Sources
Last validated 2026-08-03. Memory figures are estimates. See methodology.