Granite Vision Models by IBM Research
4 models2025–2026Up to 128k ctx
Capabilities
VisionAll models
MultimodalAll models
About
IBM's Granite Vision family of vision-language models designed for enterprise document understanding, OCR, chart/table extraction, and general image tasks. Includes models from the 3.2, 3.3, 4.0, and 4.1 generations. All released under Apache 2.0.
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
4 in view
Granite Vision 4.1 4BCurrent
Use when the workload needs 4B parameters and multimodal inputs.
2026-044B parametersmultimodal inputs
Granite 4.0 3B VisionCurrent
Use when the workload needs 128k context, 3B parameters, and multimodal inputs.
2026-04128k context3B parametersmultimodal inputs
Granite Vision 3.3 2BCurrent
Use when the workload needs 128k context, 2B parameters, and multimodal inputs.
2025-06128k context2B parametersmultimodal inputs
Granite Vision 3.2 2BCurrent
Use when the workload needs 128k context, 2B parameters, and multimodal inputs.
2025-02128k context2B parametersmultimodal inputs
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Granite Vision 4.1 4B | Use when the workload needs 4B parameters and multimodal inputs. | 2026-04 | 4B parametersmultimodal inputs | Current |
| Granite 4.0 3B Vision | Use when the workload needs 128k context, 3B parameters, and multimodal inputs. | 2026-04 | 128k context3B parametersmultimodal inputs | Current |
| Granite Vision 3.3 2B | Use when the workload needs 128k context, 2B parameters, and multimodal inputs. | 2025-06 | 128k context2B parametersmultimodal inputs | Current |
| Granite Vision 3.2 2B | Use when the workload needs 128k context, 2B parameters, and multimodal inputs. | 2025-02 | 128k context2B parametersmultimodal inputs | Current |
Release Timeline
3 release groups2026-04
2 current
Granite 4.0 3B Vision
Current128k context3B parametersmultimodal inputs
Granite Vision 4.1 4B
Current4B parametersmultimodal inputs
2025-06
1 current
Granite Vision 3.3 2B
Current128k context2B parametersmultimodal inputs
2025-02
1 current
Granite Vision 3.2 2B
Current128k context2B parametersmultimodal inputs
Specifications(4 models)
| Model | Released | Context | Parameters | Vision | Multimodal |
|---|---|---|---|---|---|
| Granite Vision 4.1 4B | 2026-04 | — | 4B | Yes | Yes |
| Granite 4.0 3B Vision | 2026-04 | 128k | 3B | Yes | Yes |
| Granite Vision 3.3 2B | 2025-06 | 128k | 2B | Yes | Yes |
| Granite Vision 3.2 2B | 2025-02 | 128k | 2B | Yes | Yes |
Frequently Asked Questions
- What is Granite Vision used for?
- Granite Vision is used for vision and multimodal work, coding, and structured outputs. The family description and listed model capabilities point to those workloads as the best fit.
- How does Granite Vision compare to Granite 4?
- Granite Vision by IBM Research is strongest where you need vision and multimodal work, while Granite 4 by IBM Research is the closest related family to check for audio. Granite Vision has 4 listed variants and reaches up to 128k context, while Granite 4 reaches up to 131k context, so compare the specs and pricing tables before choosing a production model.
- Which Granite Vision model should I use?
- If price is the main constraint, use the pricing table first because Granite Vision does not have complete provider pricing in the local data. For the most capable/latest local choice, evaluate Granite 4.0 3B Vision with 128k context and multimodal inputs.
