Cited GPU sheets × cited /models · llama.cpp
Can my Radeon run it?
Search “can RX 7900 XTX run Qwen3.6-35B” or “GLM 5.3 Flash on 16 GB Radeon.” Each pair uses the calculator already on AMDai — the same VRAM math as /calculator, the cited /gpus sheets, and the cited /models pages. Q6 / Q5 / Q4 plus a labeled Q3 estimate. Measured decode is only the official local desktop 35B llama.cpp Vulkan sweep on R9700. Everything else stays not measured yet. llama.cpp first. We do not invent a price.
How we test — Measured cells cite official amdai-bench/run@1 cards. Estimates stay labeled. Q3 is estimate-only. Q1/Q2 stay off.
Every cited pair
One cell per GPU × model. Verdicts: Fits fully on GPU, Fits with RAM offload, Needs 2 GPUs, or Does not fit. Shopper files stay Q4 and up at 4K on the calculator.
FAQ
- What does “can my Radeon run it” mean here?
- A calculator verdict on a cited GPU sheet × a cited /models page. Fits fully on GPU, fits with RAM offload, needs 2 GPUs, or does not fit. Shopper files stay Q4 and up. Q3 is a labeled estimate.
- Can my Radeon run it? · Calculator · Models
- Should I download the Q3 file?
- Q3 rows are labeled estimates from the publisher listing. This site does not recommend a Q3 as the shopper file. Q1/Q2 stay off.
- Quants · Q6 / Q5 / Q4 + Q3 estimate
If the app is not installed, packages are not published yet. There is no public download URL. Read the Stack Lab install notes . If Stack Lab is already running on this machine, open the local app .
Also: Calculator · GPUs · Models · Compare GPUs · Compare models · Speed up llama.cpp