IBM/Mixture of Experts

IBMgranite-4.0-h-tiny 6.9B

chat
6.9B
Parameters (1.5B active)
128K
Context length
9
Benchmarks
6
Quantizations
28K
HF downloads
Architecture
MoE
Released
2025-10-02
Layers
40
KV Heads
4
Head Dim
128
Family
granite

Quantization Options

QuantBitsVRAMQuality
Q4_K_M4.894.7 GBgood
Q5_K_S5.575.3 GBgood
Q5_K_M5.75.4 GBgood
Q6_K6.566.1 GBexcellent
Q8_08.57.8 GBlossless
FP161614.3 GBlossless

Select your GPU above to see speed estimates and compatibility for each quantization.

Benchmarks (9)

HumanEval83.0
IFEval81.4
MBPP80.0
BBH66.3
GPQA32.6
alpacaeval30.6
MMLU-PRO27.9
MATH23.8
MUSR16.8

Run this model

Easiest way to get starteddocs →
curl -fsSL https://ollama.com/install.sh | sh
$ollama run granite:6b-q4_k_m

Downloads and runs automatically. Add --verbose for speed stats.

Setup guide

GPUs that can run this model

At Q4_K_M quantization. Sorted by minimum VRAM.

Find the best GPU for granite-4.0-h-tiny 6.9B

Build Hardware for granite-4.0-h-tiny 6.9B

granite-4.0-h-tiny 6.9B6.9B Parameter Mixture of Experts LLM

Model Specifications

Parameters
6.9B (1.5B active)
Architecture
Mixture of Experts
Context Length
128K tokens
Capabilities
chat
Release Date
2025-10-02
Provider
IBM
Family
granite

VRAM Requirements

QuantizationBPWVRAMQuality
Q4_K_M4.894.7 GB94%
Q5_K_S5.575.3 GB96%
Q5_K_M5.75.4 GB96%
Q6_K6.566.1 GB97%
Q8_08.57.8 GB100%
FP161614.3 GB100%

Benchmark Scores

HumanEval83.0
MMLU-PRO27.9
MATH23.8
IFEval81.4
BBH66.3
GPQA32.6
MUSR16.8
MBPP80.0
alpacaeval30.6

How to Run granite-4.0-h-tiny 6.9B

Run granite-4.0-h-tiny 6.9B locally with Ollama (needs 4.7 GB VRAM at Q4_K_M):

ollama run granite:6b

Compatible GPUs (30)

GPUs that can run granite-4.0-h-tiny 6.9B at Q4_K_M quantization: