deepseek-v4-flash
Architecture mixture of experts43 LayersLicense mitMaximum context 1,024kVocabulary 129,280
At IQ4_XS it is a 127.28 GiB file, and it fits on 2 of the 18 machines we measure (the one with the least memory: PC with Ryzen AI Max+ PRO 495 and 192 GB unified). Here the file is what fills the memory: even at its maximum context of 1024k tokens, the context adds 86.00 GiB — and the same content in German takes ×1.58 the tokens of English.
Model size at IQ4_XS
Measured by Local AI Scope
Context memory at 8k tokens
Measured by Local AI Scope
What German costs versus English
Measured by Local AI Scope
Sizes of the real published files
These are the sizes of the files actually published by the author, added up when the model ships split into parts. Not an estimate from the parameter count.
| Compression | Model size |
|---|---|
IQ4_XS |
127.28 GiB |
Q3_K_M |
119.28 GiB |
Context memory
Computed from the attention pattern this model declares layer by layer, not from the generic formula. That is why the figure is often far smaller than other calculators tell you.
- 8k tokens → 0.67 GiB
- 32k tokens → 2.69 GiB
- 128k tokens → 10.75 GiB
What it costs in each language
Measured by running this model’s own tokenizer over the same content in four languages. Tokens are what you pay for in context, in memory and in your cloud bill.
| Kind of text | EN | ES | FR | DE |
|---|---|---|---|---|
| Code | — | ×1.31 | ×1.44 | ×1.60 |
| Contracts | — | ×1.37 | ×1.47 | ×1.64 |
| Business email | — | ×1.30 | ×1.38 | ×1.49 |
| Support tickets | — | ×1.43 | ×1.53 | ×1.58 |
English is the baseline: every figure is how many times more tokens the same content costs in that language. Average across the whole corpus: DE ×1.58.
Where it fits
Fits means the model, its context memory and the working headroom all fit at the stated context, leaving the system its share. 2 of the 18 machine profiles tested (IQ4_XS, 8k tokens).
If you are willing to compress it more, it does run where the table says no: at Q3_K_M (119.28 GiB), up to 32k tokens of context, on Mac Studio with M5 Max and 128 GB. Below IQ4_XS the model loses a little quality, and at Q3_K_M you can tell; that is why the table above sticks to IQ4_XS.