Granite-3.3-8B-Instruct
Enterprise 8B instruct model with 128K context and structured reasoning.
About
IBM's enterprise-focused 8B instruct model: 128K context, 12 supported languages, and structured <think>/<response> tags that separate reasoning from output.
Trained on permissively licensed and synthetic data under Apache 2.0, it's built for RAG, function calling, and code tasks inside companies that care about data governance.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Improved scores on AlpacaEval 2.0 and Arena-Hard versus Granite 3.1 per IBM; exact numbers not published in the sources reviewed.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (metrics tier, metrics-v1). Evidence: Solid enterprise adoption via watsonx.ai, Ollama, and AWS Marketplace; IBM maintains the Granite line with regular releases (4.0 already out); Apache 2.0 with documented training-data governance. Manual review: release date, 128K context, and 12-language support verified on the HF model card and AWS listing. License: Apache 2.0.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 9.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · Solid enterprise adoption via watsonx.ai, Ollama, and AWS Marketplace
- · IBM maintains the Granite line with regular releases (4.0 already out)
- · Apache 2.0 with documented training-data governance
- · manual review: release date, 128K context, and 12-language support verified on the HF model card and AWS listing