Qwen2.5-72B is the flagship open-weight model of the series, featuring 72 billion parameters, 80 layers, and a 64/8 attention head architecture, representing the pinnacle of Alibaba's open-source language model capabilities. The model supports a 128K-token context window with 8K-token generation, enabling it to analyze multiple documents and produce detailed content with exceptional accuracy.
Trained on an extended dataset of 18 trillion tokens with enhanced filtering and specialized data in mathematics and programming, Qwen2.5-72B delivers outstanding performance across a wide range of tasks. Its most remarkable feature is achieving state-of-the-art results among open-weight models while being significantly smaller than competitors. According to the technical report, the model demonstrates performance competitive with Llama-3-405B-Instruct, despite being five times smaller in size.
Distributed under the special Qwen Research License, Qwen2.5-72B is designed for projects requiring the highest quality natural language processing. The model is ideally suited for: fundamental AI research, development of cutting-edge AI products, training and fine-tuning specialized models, serving as a foundation for multimodal systems and building advanced AI agents
| Model Name | Context | Type | GPU | TPS | Status | Link |
|---|
There are no public endpoints for this model yet.
Rent your own physically dedicated instance with hourly or long-term monthly billing.
We recommend deploying private instances in the following scenarios:
| Name | vCPU | RAM, MB | Disk, GB | GPU | |||
|---|---|---|---|---|---|---|---|
| 16 | 131072 | 160 | 4 | $1.75 | Launch | ||
| 16 | 131072 | 160 | 4 | $3.23 | Launch | ||
| 16 | 131072 | 160 | 4 | $4.26 | Launch | ||
| 16 | 98304 | 160 | 3 | $4.34 | Launch | ||
| 24 | 262144 | 160 | 2 | $5.35 | Launch | ||
| 24 | 262144 | 160 | 2 | $10.40 | Launch | ||
Contact our dedicated neural networks support team at nn@immers.cloud or send your request to the sales department at sale@immers.cloud.