Will DeepSeek V4.1 Flash fit on your machine?

DeepSeek V4.1 Flash ships as 510.3 GB across 48 files. Most of that never has to be in memory at once. Set your memory below and see which build fits, and what ends up on disk. Only the byte count matters here, so two machines with the same memory get the same answer.

You will see this called a 748B model and a 510 GB download, and both are right. The experts ship packed two 4-bit weights to a byte, so 552B backbone parameters occupy 296 GB instead of the terabyte they would need at full precision. Parameter count is the wrong unit for this question. Bytes is the right one.

The lookup tables do not need to be in memory

Of the 510.3 GB, 203.1 GB is a pair of hashed n-gram lookup tables. The number looks fatal until you check how much of it gets read. From the released inference code: 48 rows per token at 264 bytes each, so 12.4 KiB per token. Next to the 4.51 GB of expert weights each token pulls in, the tables are rounding error.

So they are a latency cost, not a bandwidth cost, and they can sit on an SSD. Qwen3.8 has a similarly large static n-gram table, and on an M5 Max the difference is measurable: leaving that table on SSD instead of in memory costs 0.9% of decode speed, 40.09 against 40.47 tok/s.

What the 510.3 GB is made of

Context memory is not the constraint

890 bytes per token, so a full million-token context stays under a gigabyte. Plan around the file size.

How fast does it actually run?

One published local measurement exists as of 2026-09-10. Other figures in circulation are either hosted API throughput or the previous model, V4 Flash, which has no lookup tables and is a third of the size.