models · qwen
7B dense instruct model with 128K context and strong code/math performance for its size. Apache-2.0 open weights.
| Weight file SHA-256 | f5949308c7ab1f6e20d3c05945be7179016bc7e8b36d4cceb923ae1ba58215a8 |
|---|---|
| Method | hf-published-sha256 |
| Publisher | qwen |
|---|---|
| Origin | Recorded as a covered nation under 10 U.S.C. § 4872(f) — the PRC, Russia, Iran or North Korea. The registry did not record which. |
| Licence | Apache-2.0 |
| Parameters | 7,615,616,512 (7.62B) |
| Context window | 128K |
| Modality | text->text |
| Architecture | qwen2 |
| Layers | 28 |
| Hidden size | 3584 |
| Attention | GQA (28 heads, 4 KV) |
| Native precision | BF16 |
| Pulls recorded here | 412,014 |
| Downloads reported upstream | 9,947,372 |
| Hardware | Engine | Build measured | Decode, 1 stream | Label | Reported by |
|---|---|---|---|---|---|
| Apple M1 Max 64GB | ollama | Q4_K_M | 48.7 tok/s @ 2,048 | candidate | localmaxxing |
| Measured outside 2K–32K context — decode speed depends heavily on context, so these are not comparable with the rows above | |||||
| GeForce RTX 4070 SUPER | llama.cpp | Q4_K_M | 77.4 tok/s @ 65,536 | candidate | localmaxxing |
From local-ai-registry (MIT), commit 124959d: 2 runs across 2 machines, 0 validated — model revision and runtime pinned, launch accepted — and 2 candidate, their label for useful evidence without a reproducible-launch promise.
Registry record as of 2026-06-11. Identity, licence and parameter count are properties of a release and do not move; tier can, and the verify link above re-checks it against the live registry rather than this page.