Granite 3 Models by IBM Research
About
IBM's Granite 3.0, a family of open-source large language models (LLMs), is tailored for enterprise applications, offering a powerful blend of performance and usability 124. These models are specifically designed to accommodate various tasks and resource constraints, making them versatile for business use. Key features include exceptional performance benchmarks, comprehensive safety protocols, and support for multiple languages and programming languages 124. Released under the Apache 2.0 license, Granite 3.0 models promote transparency and community collaboration, providing enterprises with adaptable and economical AI solutions 124.
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
Use when the workload needs 128k context, 8B parameters, and tool use.
Use when the workload needs safety, 8k context, and 8B parameters.
Use when the workload needs 128k context and 8B parameters.
Use when the workload needs 128k context and 2B parameters.
Use when the workload needs 128k context and 8B parameters.
Use when the workload needs 128k context and 8B parameters.
Use when the workload needs 128k context and 2B parameters.
Use when the workload needs 128k context and 2B parameters.
Use when the workload needs 128k context and 3B parameters.
Use when the workload needs 128k context and 3B parameters.
Use when the workload needs 128k context and 1B parameters.
Use when the workload needs 128k context and 1B parameters.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Granite 3.3 8B Instruct | Use when the workload needs 128k context, 8B parameters, and tool use. | 2025-03 | 128k context8B parameterstool use | Current |
| Granite Guardian 3.0 8B | Use when the workload needs safety, 8k context, and 8B parameters. | 2025-03 | safety8k context8B parameters | Current |
| Granite 3.1 8B Instruct | Use when the workload needs 128k context and 8B parameters. | 2024-12 | 128k context8B parameters | Current |
| Granite 3.1 2B Instruct | Use when the workload needs 128k context and 2B parameters. | 2024-12 | 128k context2B parameters | Current |
| Granite 3.1 1B A400M Instruct | Use when the workload needs 128k context. | 2024-12 | 128k context | Current |
| Granite 3.1 3B A800M Instruct | Use when the workload needs 128k context. | 2024-12 | 128k context | Current |
| Granite 3 8B Instruct | Use when the workload needs 128k context and 8B parameters. | 2024-10 | 128k context8B parameters | Current |
| Granite 3 8B | Use when the workload needs 128k context and 8B parameters. | 2024-10 | 128k context8B parameters | Current |
| Granite 3 2B Instruct | Use when the workload needs 128k context and 2B parameters. | 2024-10 | 128k context2B parameters | Current |
| Granite 3 2B | Use when the workload needs 128k context and 2B parameters. | 2024-10 | 128k context2B parameters | Current |
| Granite 3 3B A800M Instruct | Use when the workload needs 128k context and 3B parameters. | 2024-10 | 128k context3B parameters | Current |
| Granite 3 2B A800M | Use when the workload needs 128k context and 3B parameters. | 2024-10 | 128k context3B parameters | Current |
| Granite 3 1B A400M Instruct | Use when the workload needs 128k context and 1B parameters. | 2024-10 | 128k context1B parameters | Current |
| Granite 3 1B A400M | Use when the workload needs 128k context and 1B parameters. | 2024-10 | 128k context1B parameters | Current |
Release Timeline
3 release groupsSpecifications(14 models)
| Model | Released | Context | Parameters | Fn Calling | Tool Use |
|---|---|---|---|---|---|
| Granite 3.3 8B Instruct | 2025-03 | 128k | 8B | Yes | Yes |
| Granite Guardian 3.0 8B | 2025-03 | 8k | 8B | No | No |
| Granite 3.1 8B Instruct | 2024-12 | 128k | 8B | No | No |
| Granite 3.1 2B Instruct | 2024-12 | 128k | 2B | No | No |
| Granite 3.1 1B A400M Instruct | 2024-12 | 128k | 1B total / 400M active | No | No |
| Granite 3.1 3B A800M Instruct | 2024-12 | 128k | 3B total / 800M active | No | No |
| Granite 3 8B Instruct | 2024-10 | 128k | 8B | No | No |
| Granite 3 8B | 2024-10 | 128k | 8B | No | No |
| Granite 3 2B Instruct | 2024-10 | 128k | 2B | No | No |
| Granite 3 2B | 2024-10 | 128k | 2B | No | No |
| Granite 3 3B A800M Instruct | 2024-10 | 128k | 3B | No | No |
| Granite 3 2B A800M | 2024-10 | 128k | 3B | No | No |
| Granite 3 1B A400M Instruct | 2024-10 | 128k | 1B | No | No |
| Granite 3 1B A400M | 2024-10 | 128k | 1B | No | No |
Available From(2 providers)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| Granite 3.3 8B Instruct | Replicate API | $0.03 | $0.25 | Serverless |
Frequently Asked Questions
- What is Granite 3 used for?
- Granite 3 is used for safety, agent workflows and tool use, and coding. The family description and listed model capabilities point to those workloads as the best fit.
- How does Granite 3 compare to Granite 4?
- Granite 3 by IBM Research is strongest where you need safety, while Granite 4 by IBM Research is the closest related family to check for audio. Granite 3 has 14 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 3 model should I use?
- For the lowest listed input price, start with Granite 3.3 8B Instruct through Replicate API at $0.03/1M input tokens. For the most capable/latest local choice, evaluate Granite 3.3 8B Instruct with 128k context and tool use and function calling.
Related Model Families
Models(14)
Granite 3.3 8B Instruct
Granite Guardian 3.0 8B
Granite 3.1 8B Instruct
Granite 3.1 2B Instruct
Granite 3.1 1B A400M Instruct
Granite 3.1 3B A800M Instruct
Granite 3 8B Instruct
Granite 3 8B
Granite 3 2B Instruct
Granite 3 2B
Granite 3 3B A800M Instruct
Granite 3 2B A800M
Granite 3 1B A400M Instruct
Granite 3 1B A400M