DeepSeek V4 Flash Flash on RTX 5090 Laptop GPU MXFP4
too tight
On an RTX 5090 Laptop GPU (24 GB), DeepSeek V4 Flash Flash in MXFP4 needs about 147,1 Go of video memory: too tight, part of the model would have to be offloaded to system memory (much slower).
Not measured yet: the figures below are estimates, computed from the models already measured on this card. Measure my card
—
Generation (estimated)
≥ 147,1 Go
Recommended VRAM (24 GB avail.)
—
Power (ref. model)
—
Electricity / M tokens · 0,25 €/kWh
No speed estimate: the model does not fit in video memory, part of it would spill into system memory and speed would collapse (often below 5 tok/s).
What about a longer context?
Context
KV cache
Estimated VRAM
Fits in 24 GB?
Estimated speed
4 096
0,3 Go
146,5 Go
no
—
8 192
0,7 Go
146,8 Go
no
—
16 384
1,3 Go
147,5 Go
no
—
32 768
2,7 Go
148,8 Go
no
—
65 536
5,4 Go
151,5 Go
no
—
131 072
10,8 Go
156,9 Go
no
—
Estimated VRAM = file + KV cache (full-attention layers, f16) + ~0,5 Go of buffers, excluding memory used by the system — hence the slightly higher “recommended VRAM”, which keeps that margin. Speed ≈ estimated speed × weights read ÷ (weights read + added KV cache).
Estimates based on real measurements of other models on this card and on the model's public data (DeepSeek). As soon as a measurement exists, this page shows the real figures.