Granite 4.2 30B

Name: Granite 4.2 30B
Author: IBM (Granite)

IBM (Granite) · Granite 4.2 · released 2026-08-07 · data checked 2026-11-01

Granite 4.2 30B is a 30B-parameter dense model from IBM (Granite) (United States), released 2026-08-07 under the Apache-2.0 license. It scores 67.0 on the Small LM Index, #11 of 34 medium models. It needs about 21.4 GB of memory at 4-bit with an 8K-token context.

Small LM Index

67.0

#11 of 34 scored medium models

Likely 65.5–68.5 · from 4 of 6 core benchmarks

Parameters

30B

Dense

Price per 1M tokens

—

No hosted API on OpenRouter; self-host

Context window

131K

tokens

Benchmarks

Each bar shows this model's score. The tick marks the best medium model on that benchmark. Sources: 8 reported by IBM (Granite). How we label sources.

MMLU-Pro Lab-reported Knowledge77.6

Class best: Qwen3.6 27B 86.2

GPQA Diamond Lab-reported Knowledge66.4

Class best: Nanbeige4.3 33B 89.5

AIME Lab-reported Math · AIME 202589.2

Class best: Ternary Bonsai 2 27B 95.8

LiveCodeBench Lab-reported Coding · LiveCodeBench v675.8

Class best: Qwen3.8 27B 90.3

Terminal-Bench Lab-reported Agentic · Terminal-Bench 2.129.2

Class best: Nanbeige4.3 33B 80.9

IFBench Lab-reported Instruction following · Prompt-level77.2

Class best: Sage-A1.5 81.6

BFCL Lab-reported Agentic · BFCL v461.4

Class best: Ternary Bonsai 2 27B 74.9

HMMT Lab-reported Math · HMMT Feb 202589.2

Not reported: estimated

What the index model expects from this model's other scores. Not a measurement.

Humanity's Last Exam≈ 15

SWE-bench Verified≈ 58

Specifications

Lab: IBM (Granite) (US · United States)
Architecture: Dense transformer
Total parameters: 30B
Active parameters: All (dense)
Size class: Medium
Context window: 131K tokens
Input modalities: text
Reasoning: Instruct (no thinking)
License: Apache-2.0
Openness: Open weights
Memory to run: ~21.4 GB at 4-bit, 8K context (more sizes)
Hugging Face: 32K downloads (30 days) · 140 likes

Weights on Hugging Face ↗

Compare with

Granite 4.2 30B vs Muse Glimmer 30B (29.6B, index 72.6)
Granite 4.2 30B vs Gemma 4 31B (30.7B, index 70.1)
Granite 4.2 30B vs Qwen3 30B-A3B Thinking 2507 (30.5B, index 59.3)
Granite 4.2 30B vs Kanana-2 30B-A3B Thinking (30.7B, index 50.7)

Ranked in

#6 in Best American open models under 100B

Sources

Something wrong here? Report it.

Run it locally

Uses ibm-granite/granite-4.2-30b-GGUF (Q4_K_M). Swap the tag for another size: BF16, Q2_K, Q3_K_L, Q3_K_M, Q3_K_S, Q4_0, Q4_1, Q4_K_M….

Ollama

ollama run hf.co/ibm-granite/granite-4.2-30b-GGUF:Q4_K_M

llama.cpp

llama-server -hf ibm-granite/granite-4.2-30b-GGUF:Q4_K_M

Then open http://localhost:8080 for a chat window.

MLX (Mac)

pip install -U mlx-lm
mlx_lm.chat --model lmstudio-community/granite-4.2-30b-MLX-4bit

vLLM (GPU server)

vllm serve ibm-granite/granite-4.2-30b

Full-precision weights, for a server GPU.

Open in LM Studio

Memory needed: weights, context cache and runtime buffers.

Quantization	4K context	32K context	131K context
16-bit F16	62 GB	70 GB	95 GB
8-bit Q8_0	34 GB	42 GB	67 GB
4-bit Q4_K_M	20 GB	28 GB	54 GB

Check it against your GPU, Mac or phone →