Steam Deck OLED · Qwen3 8B · Q4_K_M
Yes. Steam Deck OLED runs Qwen3 8B at Q4_K_M with room left at 8,192 tokens, roughly 9–16 tokens per second.
| Piece | Value | Note |
|---|---|---|
| Weights | 4.68 GiB | Qwen3-8B-Q4_K_M.gguf |
| KV cache | 1.13 GiB | 8,192 tokens |
| Engine overhead | 0.50 GiB | Flat 0.5 GiB allowance |
| 5% headroom | 0.32 GiB | On the sum above |
| Total | 6.62 GiB | Fits |
| Fast memory | 12.0 GiB | After OS reserve |
| Speed | 9–16 tok/s | Dense bandwidth range |
ollama run qwen3:8b
Steam Deck OLED · Qwen3 8B ·GGUF file · Methodology
FAQ
How was this fit calculated?
Weights 4.68 GiB, KV cache 1.13 GiB at 8,192 tokens, engine overhead 0.50 GiB, then 5% headroom, for 6.62 GiB. Fast memory available is 12.0 GiB.
What if I change the quant or the context?
This page is Q4_K_M at 8,192 tokens. A smaller quant or a shorter context lowers the total. The homepage calculator changes both without a new page.
Can I trust the tokens per second?
The range 9–16 tok/s is a bandwidth roofline, 55% to 80% efficient. This dense model reads the whole weight file each token. Measure your own box before you buy around it.
What do I type?
ollama run qwen3:8b
Is Steam Deck OLED assumed to have extra RAM?
No separate RAM pool. The chip's unified memory is the whole budget, minus the OS reserve described on the methodology page.