Configure a rig →

Granite1.3B params0.4B active (MoE)4k context

Granite 3 1B A400M locally

Granite 3 1B A400M is an open-weight LLM from the Granite family with 1.3B parameters. Main use: chat, RAG and general assistance. Detected minimum hardware: GTX 1650 (4 GB).

Check on my GPU →Hugging Face ↗

Technical facts

Parameters1.3B

Max context4k

Q4_K_M0.8 GB

Q5_K_M1.0 GB

Q81.5 GB

FP162.9 GB

FamilyGranite

Last sync2026-05-12

Available quantizations

GGUF weights

Q4_K_M

0.8GB

Acceptable. Good compromise when VRAM is limited.

Q5_K_M

1.0GB

Good quality. Sweet spot for size and precision.

Q8

1.5GB

Near-FP16 quality. Comfortable for production.

FP16

2.9GB

Reference precision. Maximum quality, doubled VRAM.

Compatible GPUs

12 single-GPU

GPUs that can run Granite 3 1B A400M on a single card, ranked by VRAM headroom.

2.9 / 4 GBcomfortable · FP16

2.9 / 6 GBcomfortable · FP16

2.9 / 6 GBcomfortable · FP16

2.9 / 6 GBcomfortable · FP16

2.9 / 6 GBcomfortable · FP16

2.9 / 6 GBcomfortable · FP16

2.9 / 6 GBcomfortable · FP16

6 GB · Arc Alchemist

2.9 / 6 GBcomfortable · FP16

2.9 / 8 GBcomfortable · FP16

2.9 / 8 GBcomfortable · FP16

2.9 / 8 GBcomfortable · FP16

2.9 / 8 GBcomfortable · FP16

Recommended multi-GPU rigs

2x / 4x consumer GPUs

For Granite 3 1B A400M at higher quantization or with more context, a multi-GPU rig gives more headroom.

2.9 / 8 GBcomfortable · FP16

2× GTX 1060 6GB

12 GB · GTX 10

2.9 / 12 GBcomfortable · FP16

12 GB · GTX 16

2.9 / 12 GBcomfortable · FP16

2× GTX 1660 Super

12 GB · GTX 16

2.9 / 12 GBcomfortable · FP16

2× GTX 1660 Ti

12 GB · GTX 16

2.9 / 12 GBcomfortable · FP16

2× RTX 2060 6GB

12 GB · RTX 20

2.9 / 12 GBcomfortable · FP16

2× RTX 3050 6GB

12 GB · RTX 30

2.9 / 12 GBcomfortable · FP16

12 GB · Arc Alchemist

2.9 / 12 GBcomfortable · FP16

Recommended rig

2× GTX 1650

Granite 3 1B A400M with Ubuntu, vLLM, Open WebUI and the model already downloaded.

Similar models

VRAM estimates: parameters x bits/8 plus margin. Real performance varies by engine, context length and batch size.
sync: 2026-05-12