DeepSeek-R1-0528-Qwen3-8B

Good fit for

People experimenting with a smaller local reasoning model rather than the full DeepSeek-R1 model.

Why consider it

The current runtime tag identifies the 0528 Qwen3-based 8B distillation, making the lineage explicit.

Main limitations

The tag currently resolves to a Qwen3-based 8B model, not the older Llama-based 8B distillation. Download size is not working-memory demand. Do not transfer full-model benchmark claims to this distilled variant.

Cost and access

No hosted subscription or per-token fee for this local package. Supply your own hardware; electricity and setup are not free. The catalog lists MIT licensing; inspect the exact weights and upstream terms before redistribution.

Getting started

Install a compatible Ollama version, then run: ollama run deepseek-r1:8b. The API filter refers to the runtime-provided local API, not a bundled hosted endpoint.