granite3-moe:latest

Upstream reference: Granite 3.0 1b-a400m Instruct

Publisher: IBM Granite Team

IBM describes this instruction-tuned sparse mixture-of-experts text model for assistants and business-oriented document workflows.

Publisher-described context, separate from Sigma5C observations and recorded catalog statistics.

Published upstream specifications

Total parameters

1.3 billion Source

Active parameters

400 million Source

Published sequence length

4096 tokens Source

Upstream license label

Apache-2.0 Source

Publisher-described uses

• Text extraction, classification, summarization and retrieval-assisted question answering. Source

Reference limitations

• The later Granite 3.1 long-context figures must not be assigned to this 3.0 checkpoint. Source

A recorded tag and an upstream card association do not verify the deployed checkpoint, quantization, configured context or current hosted availability.

Publisher sources

• Granite 3.0 1b-a400m Instruct official documentation — IBM Granite Team. Checked 2026-11-02.

Links

Recorded references do not establish downloadable weights, licensing or current availability.

• IBM — official site
• Ollama
• Provider page
• Hugging Face
• Official model announcement / documentation

Where granite3-moe:latest comes from

Named in 1 observation page

• Xeon vs AMD workstation — CPU inference head-to-head · V1