Qwen 2.5 7B instruct
About this model
Qwen2.5 7B is the latest series of Qwen large language models. Qwen2.5 brings the following improvements upon Qwen2: - Significantly more knowledge and has greatly improved capabilities in coding and mathematics, thanks to our specialized expert models in these domains. - Significant improvements in instruction following, generating long texts (over 8K tokens), understanding structured data (e.g, tables), and generating structured outputs especially JSON. More resilient to the diversity of system prompts, enhancing role-play implementation and condition-setting for chatbots. - Long-context Support up to 128K tokens and can generate up to 8K tokens. - Multilingual support for over 29 languages, including Chinese, English, French, Spanish, Portuguese, German, Italian, Russian, Japanese, Korean, Vietnamese, Thai, Arabic, and more. Usage of this model is subject to [Tongyi Qianwen LICENSE AGREEMENT](https://huggingface.co/Qwen/Qwen1.5-110B-Chat/blob/main/LICENSE).
Performance Tier
Qwen 2.5 7B instruct is a compact model from Qwen : optimized for speed and affordability.
Small, fast, and affordable. Optimized for speed and low cost, great for high-volume or simple tasks.
Pricing
This model is included in Elosia plans| Type | per 1M tokens |
|---|---|
| Input (prompt) | $0.040 |
| Output (completion) | $0.100 |
Capabilities
Benchmarks
Recommended Use Cases
Strengths
- Strong performance for a 7B parameter model
- Excellent multilingual support (29+ languages including French)
- Open-weight with Apache 2.0 license
- Very low cost ideal for high-volume deployments
Limitations
- Limited complex reasoning due to small model size
- Not suitable for advanced coding or research tasks
Resources
This model may use your data for training
