Qwen2.5-7B
Qwen2.5-7B is the pretrained base 7B language model released with Qwen2.5 on September 19, 2024. It is separate from the instruction-tuned, Coder, Math and vision variants.
Benchmark scores
Access
Open-weight release under the exact repository’s Apache 2.0 licence. The card advertises 131,072-token context; no runtime test, model-specific output cap or hosted API/GA availability is verified. The card recommends post-training rather than using the base model directly for conversations. Qwen-Plus and Qwen-Turbo APIs are separate identities.