IBM Granite 4.2 — local AI model by IBM
Written by Jakub Rusinowski · Last updated
IBM's first family of dense decoder-only reasoning models, released August 2026 in 3B, 8B and 30B. Granite 4.2 drops the hybrid Mamba-2 MoE design of Granite 4.0 and continues the dense line Granite 4.1 started. Each size is pretrained from scratch on roughly 15T tokens with a five-phase context extension to 512K, then RL-post-trained for tool calling, RAG, structured JSON and an explicit thinking mode. All three are Apache 2.0.
Variants
The smallest IBM Granite 4.2 variant needs about 3 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache.
| Model | VRAM at Q4 | VRAM | Context | Run it |
|---|---|---|---|---|
| Granite 4.2 3B → 3B | ~2.6 GB | 131,072 | ollama run granite4.2:3b | |
| Granite 4.2 8B → 8B | ~5.6 GB | 131,072 | ollama run granite4.2:8b | |
| Granite 4.2 30B → 30B | ~18.9 GB | 131,072 | ollama run granite4.2:30b |
Memory is quantized weights plus overhead at Q4_K_M, from the same engine as the GPU & VRAM checker.
How to run IBM Granite 4.2 locally
Install Ollama, then pull the tag.
ollama run granite4.2:3bPick a size above for its own VRAM figure, speed estimate and install command.
Licence
Commercial use permitted. No usage restrictions beyond attribution.
Applies to: Granite 4.2 3B, Granite 4.2 8B, Granite 4.2 30BRecommended GPU
The cheapest catalogued GPU that runs IBM Granite 4.2 locally (min 3 GB VRAM) is the Intel Arc B570 (10 GB).
Can I run IBM Granite 4.2 on my GPU?
- IBM Granite 4.2 on AMD Radeon RX 7900 XT
- IBM Granite 4.2 on AMD Radeon RX 7900 XTX
- IBM Granite 4.2 on AMD Radeon RX 9060 XT 8GB
- IBM Granite 4.2 on Apple M1 Pro
- IBM Granite 4.2 on Apple M2 Pro
- IBM Granite 4.2 on Apple M3 Pro
- IBM Granite 4.2 on Apple M4
- IBM Granite 4.2 on Apple M5
- IBM Granite 4.2 on Intel Arc B570
- IBM Granite 4.2 on Intel Arc B580
- IBM Granite 4.2 on NVIDIA GeForce RTX 3060 (12GB)
- IBM Granite 4.2 on NVIDIA GeForce RTX 3070
- IBM Granite 4.2 on NVIDIA GeForce RTX 3070 Ti
- IBM Granite 4.2 on NVIDIA GeForce RTX 3080 (10GB)
- IBM Granite 4.2 on NVIDIA GeForce RTX 3090
- IBM Granite 4.2 on NVIDIA GeForce RTX 4060
- IBM Granite 4.2 on NVIDIA GeForce RTX 4070
- IBM Granite 4.2 on NVIDIA GeForce RTX 4070 Super
- IBM Granite 4.2 on NVIDIA GeForce RTX 4070 Ti
- IBM Granite 4.2 on NVIDIA GeForce RTX 4090
- IBM Granite 4.2 on NVIDIA GeForce RTX 5060
- IBM Granite 4.2 on NVIDIA GeForce RTX 5060 Ti 8GB
- IBM Granite 4.2 on NVIDIA GeForce RTX 5070
- IBM Granite 4.2 on NVIDIA GeForce RTX 5090
IBM Granite 4.2 — frequently asked questions
How much VRAM does IBM Granite 4.2 need?
IBM Granite 4.2 needs about 3 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: Granite 4.2 3B (3 GB, Q4_K_M); Granite 4.2 8B (6 GB, Q4_K_M); Granite 4.2 30B (19 GB, Q4_K_M). On Apple Silicon, unified memory counts toward this requirement.
Can I run IBM Granite 4.2 on an RTX 4090 (24 GB)?
Yes — IBM Granite 4.2 runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
What quantization should I use for IBM Granite 4.2?
Q4_K_M is the best balance of quality and VRAM for IBM Granite 4.2 in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
How do I run IBM Granite 4.2 with Ollama?
Install Ollama, then run: ollama run granite4.2:3b. This downloads IBM Granite 4.2 and starts a local, OpenAI-compatible endpoint — no internet connection is needed after the initial download.