Qwen3.8 27B

Good fit for

Developers connecting a local multimodal model to a coding harness or document workflow.

Why consider it

The exact provider card documents native image/video understanding, thinking control and a 262,144-token native context.

Main limitations

The 1M context figure requires extension or a separate hosted setup, not default local inference. Local weights do not include a browser, search service or code executor. Research use requires supplied sources or external tools. Short prompts can also be affected by long-context scaling.

Cost and access

No subscription or per-token hosted fee for local inference within permitted uses. Hardware, electricity and setup are separate. Apache 2.0 is confirmed by the exact provider card.

Getting started

Install a current compatible Ollama version, review the exact license and run: ollama run qwen3.8:27b. API refers to the local runtime endpoint, not a hosted API included with the weights.