Granite 4.0
IBMIBM's enterprise foundation-model family, released October 2025 as hybrid Mamba-2/transformer models from micro (3B) through h-small (32B total, 9B active MoE). The line targets efficient long-context inference on constrained hardware and enterprise agent workloads, and ships under Apache-2.0 with ISO 42001 certification and cryptographically signed checkpoints.
The Granite 4.0 mainline that granite-code-instruct's deprecation banner points to. Apache-2.0 weights with the SFT data characterized but not released and no pretraining pipeline, so open weights rather than open source. Adoption is summed across the four declared 4.0 SKUs; more exist (base variants, nano), which shifts reach without changing the reading.
Openness
3 high confidence- weights
- open(Apache-2.0, ungated on HF across the 4.0 line)
- data
- closed(SFT data characterized in the card (permissive public + internal synthetic + human-curated) but neither the pretraining corpus nor the mixtures are released)
- code
- partial(ibm-granite/granite-4.0-language-models is docs/usage, no pretraining pipeline)
- license
- Apache-2.0(OSI)
Granite 4.0 ships Apache-2.0 weights, ungated across every SKU, but IBM only describes its training data in broad categories -- public data, internal synthetic data, human-curated data -- without releasing it or specifying the mix, and no pretraining pipeline is published. That combination of open weights with closed data and code keeps Granite at open weights rather than a fuller open release.
- https://api.github.com/repos/ibm-granite/granite-4.0-language-models recorded 2026-09-01
`license: Apache-2.0`, 210 stars; the 4.0 release repository, which carries no pretraining pipeline — what holds `code` at partial.
- https://huggingface.co/api/models/ibm-granite/granite-4.0-h-small recorded 2026-10-07
Model API for ibm-granite/granite-4.0-h-small: `gated: false`, `private: false`, cardData.license `apache-2.0`, safetensors weights listed.
- https://huggingface.co/ibm-granite/granite-4.0-h-small/raw/main/README.md recorded 2026-10-07
Card Training Data section: "our SFT data is largely comprised of three key sources: (1) publicly available datasets with permissive license, (2) internal synthetic data targeting specific capabilities, and (3) a select set of human-curated data" — described in categories, not released. Infrastructure section names the GB200 cluster; no training pipeline links.
Adoption
3 high confidenceAdoption is measured as Hugging Face downloads summed across the four declared Granite 4.0 artifacts: h-tiny, micro, h-small and h-micro. The smaller, on-device-sized SKUs account for most of that reach.
- https://huggingface.co/api/models/ibm-granite/granite-4.0-h-micro recorded 2026-09-01
20,110 downloads in the trailing 30 days for ibm-granite/granite-4.0-h-micro
- https://huggingface.co/api/models/ibm-granite/granite-4.0-h-small recorded 2026-09-01
25,719 downloads in the trailing 30 days for ibm-granite/granite-4.0-h-small
- https://huggingface.co/api/models/ibm-granite/granite-4.0-h-tiny recorded 2026-09-01
88,627 downloads in the trailing 30 days for ibm-granite/granite-4.0-h-tiny
- https://huggingface.co/api/models/ibm-granite/granite-4.0-micro recorded 2026-09-01
60,053 downloads in the trailing 30 days for ibm-granite/granite-4.0-micro
Capability
3 medium confidenceMid-tier: a capable efficiency-focused line, competitive within the small-to-mid open-weight class and well short of the open-weight frontier the category's 4s and 5s hold (olmo, deepseek, qwen, kimi). Sits level with llama, whose own reading is taken on Llama 4 Maverick — the same non-reasoning open-weight mid-field.
- https://huggingface.co/ibm-granite/granite-4.0-h-small/raw/main/README.md recorded 2026-09-01
The h-small card's evaluation tables benchmark the 4.0 line against similar-size open-weight peers on instruction-following and enterprise task suites, positioning it as competitive in its size class rather than frontier.
Verified 2026-09-27, except some scores, still awaiting re-confirmation