Skip to content
← Models
Open weightsIBM/10 trustMetrics-checked

Granite Vision 3.1

Lightweight open-source document-understanding vision model from IBM, released January 2025.

Visit official page →

About

Granite Vision 3.1 is IBM's lightweight multimodal model for enterprise document intelligence, released January 31, 2025 under Apache 2.0. It combines a vision encoder with Granite 3.1-2B-Instruct and is tuned for reading tables and charts, OCR, and answering questions grounded in document content.

IBM's published numbers put it at or near the top of its size class on DocVQA (0.88), ChartQA (0.86), TextVQA (0.76), and OCRBench (0.75), matching or beating larger rivals in the 2B tier (vendor-reported). Like all Granite models it ships with enterprise-friendly governance, including dataset disclosure and indemnity protections through IBM's commercial offerings.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

January 31, 2025

Benchmarks

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

Use cases

document-understandingtable-and-chart-qaocrenterprise-automation

Trust score — 7/10

/10 trustMetrics-checked

Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 114; Trending score 74/100 from researcher assessment.

More in the granite-vision family

Sources