Prefill and token generation speed
Qwen3.8 27B on RTX 4090: prefill and generation
Watch Qwen3.8 27B process input and respond on an RTX 4090 24GB. Compare with an RTX 3090 to see how waiting time differs despite the same 24GB memory capacity.
Benchmark-based estimate · 4K input · one user
Last updated
Change device or model
Expand settings Settings4-bit · 4K · Baseline decodeCollapse
Reading this result
The 4090 also has 24GB of VRAM, so distinguish capacity from processing speed. The right comparison depends on whether you need a larger model or faster execution of one that already fits.
- Required / available memory
- 17.6GB / 22.5GB
- Runtime
- llama.cpp CUDA · Q4_K_M
Compare this model side by side
When comparing with a 3090, check document prefill as well as short chat. These are LLM results; the same ratio does not establish image or video generation performance.
Check configuration
RTX 4090 24GB · Memory 24GB
Qwen3.8 27B Q4_K_M configuration. Check that the retailer lists the same memory option.
What changes with a longer document?Expand input length, waiting time, speed and memory
Detailed analysis and all 61 hardware configurations
Popular configurations
Speed simulations by device and model
Qwen3.8 27B on Mac mini M4 32GB: response speed
5.1 ~ 5.7 tok/s · 16.4 ~ 35.5 s
Qwen3.8 27B: is a Mac mini M4 Pro 48GB worth it?
12.3 ~ 13.7 tok/s · 7.8 ~ 14.3 s
Qwen3.8 27B on MacBook Pro M5 Pro 64GB
14.7 ~ 17.1 tok/s · 8.6 to 10.0 seconds
Qwen3.8 27B on Mac Studio M4 Max 128GB
27.9 ~ 60.4 tok/s · 5.8 to 10.5 seconds