granite-4.2-3b
Official · BF16 · safetensors
Runtime certified: SGLang Transformers vLLM
| Precision / quantization | Format | Bits | Publisher trust | Deployment state |
|---|---|---|---|---|
| BF16 | safetensors | 16 | Official | Runtime certified |
IBM Granite · General · Granite 4.2
Granite 4.2 3B: Specs, GPU requirements, hosting & deployment
Granite 4.2 3B is a current General release in the IBM Granite family, certified from its immutable official source revision.
Total parameters3.0B
Active parametersDense model
Context window131,072 tokens
Release statusCurrent · Current
ArchitectureGraniteForCausalLM
ModalitiesText
LicenseApache License 2.0
Deployment stateDeployable with certified configuration
Commercial stateCommercial Use Allowed
Release date2026-08-25
Architecture, strengths and use cases
Granite 4.2 3B is an official self-hostable IBM release in the IBM Granite General track. It uses GraniteForCausalLM, carries 3B, supports a verified context window of 131,072 tokens, and exposes only source-certified checkpoints and runtime combinations. Its release, license and deployment facts are tied to the official repository revision rather than inferred from the family name.
Granite 4.2 3B uses GraniteForCausalLM with 3B. The stored architecture profile records 40 layers, hidden width 2560, 40 attention heads and a 131,072 token context ceiling where the official configuration publishes those fields.
Verified GraniteForCausalLM architecture
131,072 token certified context
Official BF16 checkpoint coverage
Private LLM inference
Agentic workflows
Long-context document processing
Certified checkpoints and quantizations
Only exact, source-certified variants attached to this release are shown.
Recommended hardware
Use only the listed transformers, vllm, sglang runtime configuration with the exact checkpoint precision. The recommendation engine checks weight memory, KV cache, runtime overhead, safety headroom, GPU architecture and topology before ranking rentable and reference hardware.
Estimated weight memory5.76 GB
KV cache0.63 GB
Runtime overhead5.87 GB
Total with safety margin15.19 GB
Recommended best value
1× NVIDIA A100 SXM 80GB
80 GB total VRAM · nvlink
64.81 GB headroom · Available
USD 915 / monthLowest-cost viable
1× NVIDIA L4 24GB
24 GB total VRAM · pcie
8.81 GB headroom · Available
USD 250 / monthHighest performance
1× NVIDIA H200 SXM 141GB
141 GB total VRAM · nvswitch
125.81 GB headroom · Limited capacity
USD 2,345 / monthLicense and self-hosting considerations
The full 3B resident weight set must fit across the selected GPUs. Context and concurrency add KV-cache memory; MoE active parameters do not replace resident-weight memory.
Granite 4.2 3B is published under Apache License 2.0. The release-scoped official license link and verification date are retained with the catalogue record.
Commercial Use Allowed. The official license permits commercial use; its obligations still apply.
Alternatives and sibling variants
Granite 4.2 30B, Granite 4.2 8B, Granite 4.1 30B
Release timeline
ibm/granite/granite-4-2-3b
Source details
Source revision: dd9a1fe61833ecf5e9963237b60da9469bf3d5cf750bd93cde2c2997bb74da98
Frequently asked questions
How much VRAM does Granite 4.2 3B require?
The exact estimate depends on checkpoint precision, context and concurrency; the page recommendation uses the certified memory engine and compatible runtime configurations.
Which runtimes support Granite 4.2 3B?
Certified upstream runtime evidence is available for transformers, vllm, sglang.
Can Granite 4.2 3B be used commercially?
The stored commercial classification is Commercial Use Allowed. Consult the official Apache License 2.0 terms for the final obligations.
Official references
Last verified: 2026-08-26