Skip to content
← Models
Open weightsIBM/10 trustVerified
Granite-3.2-2B-Instruct logo

Granite-3.2-2B-Instruct

2B enterprise instruct model with toggleable reasoning.

Visit official page →

About

Granite-3.2-2B-Instruct is IBM's 2B instruction model, released February 26, 2025 as part of the Granite 3.2 family. Its headline feature is controllable chain-of-thought reasoning that can be switched on for hard problems and off for cheap inference, trained with Thought Preference Optimization reinforcement learning.

The model supports 12 languages, long-context summarization and QA, RAG, and function calling, all under Apache 2.0. It is available on Hugging Face, watsonx.ai, Ollama, Replicate, and LM Studio.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

February 26, 2025

Benchmarks

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

enterprise-chatragsummarizationfunction-calling

Trust score — 8/10

/10 trustVerified

Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 125; Trending score 80/100 from researcher assessment.

More in the granite family

Sources