MODEL SIGNAL · QWEN
Qwen2.5-7B-Instruct
An open‑weight **instruction‑tuned 7B language model** in the Qwen2.5 series, optimized for **instruction following**, **long‑context text generation**, **structured output (e.g., JSON)**, **coding**, **mathematics**, and **multilingual general‑purpose use**.
CATEGORYGeneral
CONTEXT131072
RELEASEDSeptember 19, 2024
Key Features
- Instruction‑tuned Qwen2.5 causal language model
- ~7.6B parameters (7.61B total; 6.53B non‑embedding)
- Long‑context support up to 131,072 tokens (128K) with generation up to 8K tokens
- Config.json default context length 32,768 tokens
- Multilingual support for 29+ languages
- Improved instruction following over Qwen2
- Significantly improved coding and mathematics capabilities via specialized expert models
- Enhanced long‑text generation, structured data understanding (e.g., tables), and structured outputs (especially JSON)
Read the Model Signal report →
The designed Model Signal report for Qwen2.5-7B-Instruct is still in review. It will publish here after the fact and quality gates clear.