Skip to content
NEWLive updates feed from Hugging Face, arXiv and GitHub  See what changed →

Qwen · qwen

Qwen-72B

Qwen-72B by Qwen: a 72.3B-parameter dense open-weight model under the other license.

otherDense33K contextAuto-imported · not yet reviewed

4,169,684 downloads · 362 likes · Model card · synced 2026-09-19

Specifications

Parameters
72.3B
Architecture
dense · qwen
Layers
80
Hidden size
8,192
KV heads · head dim
64 · 128
Vocabulary
152,064
Native dtype
Context window
32,768 tokens
Released
2023-11-26
Last modified on Hub
2024-10-09

Features & licensing

License
other
Commercial use
No
Modalities
text
Tool / function calling
No
Reasoning mode
No
Languages
ZH, EN
Quantised variants on Hub
AWQ, GGUF
Pipeline
text-generation
transformerssafetensorsqwentext-generationcustom_codezhen

Task fit

Editorial scores (0–100) used by the recommendation engine.

  • Chat70
  • RAG / Q&A70
  • Code70
  • Summarisation70
  • Extraction70
  • Agents70

Hardware to run Qwen-72B

Weights need about 159 GB at FP16, 84 GB at INT8 and 48 GB at INT4. The KV cache adds roughly 2621 MB per 1,000 tokens per request. For a reference workload of 5 requests per second with 1,500 input and 300 output tokens, total GPU memory is around 261.1 GB, which fits on 4 × NVIDIA A100 (80 GB).

GPU configurations at the reference workload
ConfigurationVRAMUtilisationEst. first tokenCloud / month
6 × NVIDIA L40S (48 GB)288 GB91 %~0.2 s$8,322
4 × NVIDIA A100 (80 GB)recommended320 GB82 %~0.1 s$9,344
4 × NVIDIA H100 (80 GB)320 GB82 %~0.0 s$13,140

Estimates only. Run an assessment for your own traffic.

Auto-imported from Hugging Face. Architecture, license and feature data are synced from the model repository; the task-fit scores below are provisional defaults until a curator reviews them.