Granite 3.1 8B Instruct

Granite is IBM's open model family, built specifically for enterprise workflows. Granite 3.1 is the long-context refresh of the 3.0 release: same 8B dense architecture, expanded to a 128K context window through continued pre-training on long-document data.

Enterprise positioning

IBM ships Granite models alongside detailed model cards that include training-data provenance, governance documentation, and indemnification when used through watsonx.ai. For regulated industries this matters more than benchmark numbers — Granite is often chosen because the paperwork is straightforward, not because it tops a leaderboard.

What it's good at

Business document understanding, structured-output extraction, function calling, summarization, and English-language reasoning at the 8B scale. Granite-Code variants are popular as the LLM backbone in IBM's developer tools (watsonx Code Assistant).

What to watch for

It's a workmanlike model rather than a flashy one. Don't expect benchmark-topping performance against Qwen2.5 7B or Llama 3.1 8B. The value is in the surrounding documentation and the predictable behavior, not the raw capability.

License

Apache 2.0 — fully permissive.