FastFlowLM/GPT-OSS-20B-NPU2: what it needs to run
FastFlowLM/GPT-OSS-20B-NPU2 is a quantization of openai/gpt-oss-20b. Its configuration matches gpt-oss-20b's on every field that decides memory, so gpt-oss-20b's requirements are this repository's.
Answered by
Geometrypublished data. Both configurations are published files, read at the revisions named below and compared field by field. Not a test that the model loads.gpt-oss-20b (openai/gpt-oss-20b) — every compared field is stated by both and equal.
gpt-oss-20b: weight-file size by format
| Format | Size | Range | Basis |
|---|---|---|---|
| FP16 | 41.8 GB | 40.7 GB – 42.0 GB | reconstructed size |
| Q8_0 | 22.2 GB | 21.6 GB – 22.3 GB | reconstructed size |
| Q6_K | 17.2 GB | 16.7 GB – 17.3 GB | reconstructed size |
| Q5_K_M | 14.9 GB | 14.4 GB – 15.0 GB | reconstructed size |
| Q5_0 | 14.5 GB | 14.0 GB – 14.5 GB | reconstructed size |
| Q4_K_M | 12.8 GB | 12.3 GB – 12.9 GB | reconstructed size |
| Q4_0 | 11.9 GB | 11.4 GB – 12.1 GB | reconstructed size |
| Q3_K_M | 10.5 GB | 9.74 GB – 11.2 GB | reconstructed size |
| Q2_K | 7.81 GB | 7.33 GB – 8.13 GB | reconstructed size |
These are gpt-oss-20b’s sizes. A model with the same geometry quantizes to the same size, to within any difference in parameter count noted above. Which devices hold each, and how fast they run it, is on gpt-oss-20b’s page.
What was read, and where
- FastFlowLM/GPT-OSS-20B-NPU2 at revision 5c6a9c780204 is tagged by its publisher as a quantization of openai/gpt-oss-20b.
- FastFlowLM/GPT-OSS-20B-NPU2/config.json at 5c6a9c780204 was compared with openai/gpt-oss-20b/config.json at 6cee5e81ee83, the revision the catalogue pins: layers and their pattern, widths, head counts, experts, latent and recurrent dimensions, vocabulary, tied embeddings and any vision or audio tower.
- Licence, as tagged
- apache-2.0
- Parameters in its safetensors index
- none published
- Task, as tagged
- text-generation
- Last changed on Hugging Face
- 2026-08-04
Publishing this model? A badge for its card
[](https://llmbottleneck.com/hf/FastFlowLM/GPT-OSS-20B-NPU2)
It states the memory to run the model at Q4_K_M with 8,192 tokens of context — built on this repository’s own Q4_K_M file when it publishes one — and links back to this page, where the evidence is.
Resolve another repository
The same answer as JSON: GET /v1/resolve?repo=FastFlowLM/GPT-OSS-20B-NPU2, with a free key from /keys.