RunsOnMy ModelsGPUs & MacsBy VRAMBest GPUCan I run…?OllamaStart hereAPI pricesVRAM calculatorNew

Can Qwen AgentWorld 35B A3B run on the RX 7900 XT?

Yes. At 8K context the best fit is Q3_K_M, using about 16.6 GB of the 19.0 GB usable on the RX 7900 XT. Estimated speed ceiling: ~554 tokens/sec.

Better quality with offload: 4-bit (Q4_K_M)

Runs with offload 4-bit (Q4_K_M) at 8K context: all attention layers plus the experts of 33 of 40 MoE layers on the RX 7900 XT (about 17.4 GB), and the experts of the other 7 layers in system RAM (about 3.4 GB). Needs a PC with at least 32 GB of system RAM. Estimated speed: 116–164 tokens/sec (usable for chat), versus a bandwidth ceiling of ~418 tok/s if it fit fully in VRAM.

llama-server -hf unsloth/Qwen-AgentWorld-35B-A3B-GGUF:Q4_K_M -c 8192 -ngl 999 --n-cpu-moe 7

Estimate, not a benchmark. The command downloads the GGUF from Hugging Face. The speed range spans dual-channel DDR4-3200 to DDR5-6000 system RAM; your CPU, RAM speed and llama.cpp version move it. If llama.cpp runs out of GPU memory, raise --n-cpu-moe by 1-2; if VRAM is left over, lower it for more speed. Recent llama.cpp builds can also pick the split automatically. How we estimate offload.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K)13.1 GB Fits13.1 GB Fits13.6 GB Fits15.5 GB Fits711
3-bit (Q3_K_M) •16.5 GB Fits16.6 GB Fits17.0 GB Fits18.9 GB Tight554
4-bit (Q4_K_M) •21.6 GB No21.6 GB No22.1 GB No24.0 GB No—
5-bit (Q5_K_M) •25.6 GB No25.7 GB No26.1 GB No28.0 GB No—
6-bit (Q6_K) •28.2 GB No28.3 GB No28.8 GB No30.7 GB No—
8-bit (Q8_0) •35.3 GB No35.4 GB No35.9 GB No37.7 GB No—
16-bit (F16/BF16) •65.6 GB No65.6 GB No66.1 GB No68.0 GB No—

• real GGUF file size from unsloth/Qwen-AgentWorld-35B-A3B-GGUF. Speed is a bandwidth ceiling (800 GB/s), not a benchmark. Methodology.

What fits instead

"Runs" and "fits" mean the whole model in GPU memory at 8K context (4-bit or better for the model pick). Prices are US launch MSRPs, not today's street prices. "Popular" means the 60 most-downloaded models we track.

More

Qwen AgentWorld 35B A3B on every GPU and Mac · Every model that runs on the RX 7900 XT · Other models that fit 24 GB VRAM · Try other settings in the calculator

Guides: Best GPU for local LLMs by budget · Can I run gpt-oss, Llama 70B or Qwen3? · Mac vs PC for local LLMs

Data: Hugging Face Hub (Qwen/Qwen-AgentWorld-35B-A3B), AMD specs. Model data last checked 2026-11-18; refreshed hourly.