IBM Granite Models are a family of enterprise-ready, open foundation models that support multilingual capabilities, coding, retrieval-augmented generation (RAG), tool use, thinking and structured JSON output. Released under Apache 2.0 license.
IBM Granite Models are a family of enterprise-ready, open foundation models that support multilingual capabilities, coding, retrieval-augmented generation (RAG), tool use, and structured JSON output. Released under Apache 2.0 license.
The IBM Granite 1B and 3B models are long-context mixture of experts (MoE) Granite models from IBM designed for low latency usage.
IBM Granite 2B and 8B models are 128K context length language models that have been fine-tuned for improved reasoning and instruction-following capabilities.
The IBM Granite 2B and 8B models are text-only dense LLMs trained on over 12 trillion tokens of data, demonstrated significant improvements over their predecessors in performance and speed in IBM’s initial testing.
The IBM Granite 2B and 8B models are designed to support tool-based use cases and support for retrieval augmented generation (RAG), streamlining code generation, translation and bug fixing.
The IBM Granite 1B and 3B models are the first mixture of experts (MoE) Granite models from IBM designed for low latency usage.
Granite-3.2 is a family of long-context AI models from IBM Granite fine-tuned for thinking capabilities.
The IBM Granite Embedding 30M and 278M models models are text-only dense biencoder embedding models, with 30M available in English only and 278M serving multilingual use cases.
The IBM Granite Guardian 3.0 2B and 8B models are designed to detect risks in prompts and/or responses.
UNCENSORED (Abliterated + SecOps LoRA) — IBM Granite 4.2 Uncensored (8B Compact & 30B Flagship 128K Editions: :8b, :30b-q3, :30b, :30b-q6, :30b-q8) — Trained & Orthogonally Abliterated by Erik Tonon.
Uncensored IBM Granite 4.2 8B via Heretic: 12/100 refusals (from 99), KL 0.080. Apache-2.0, multilingual.
Parable is IBM Granite 4.1 fine-tuned on Claude Fable 5 and GPT-5.5 agent traces. Adds think reasoning to Granite. Sibling: Qwen3 line at parable/qwen3-fable
Reduced-refusal IBM Granite 4.2 3B via Heretic: 27/100 refusals (from 98), KL 0.085. Apache-2.0, multilingual.
Context-optimized variant of IBM's Granite 4.2 3B model, tuned for lightweight tool-use on GPU-constrained hosts.
Context-optimized variant of IBM's Granite 4.1 3B model, tuned for lightweight tool-use on GPU-constrained hosts.
a small local decision model for logistics operations Built on IBM Granite 3.1 1B
Granite 4 features improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.
Granite Docling is a multimodal Image-Text-to-Text model engineered for efficient document conversion.