llmrun
IBMGraniteGraniteMoeHybridForCausalLM

Granite 4.0 1B — Hardware Requirements & GPU Compatibility

ChatFunctions

Granite 4.0 1B serves as IBM's 1.6-billion-parameter instruct model within the Granite 4.0 Nano family, having been finetuned from Granite-4.0-1B-Base using supervised methods, reinforcement learning, and model merging techniques. The model focuses on general chat and assistant tasks while offering improved tool calling, fill-in-the-middle code completion, and support for English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, and Chinese. It employs a hybrid architecture (GraniteMoeHybrid as defined in the model config). At this size, it runs on nearly any consumer GPU, laptop, or CPU—even without quantization. The context window spans 131,072 tokens. Released under the Apache 2.0 license, it allows unrestricted commercial and research use. The Hub repository appeared in October 2025, with a release date of October 28, 2025 noted on the card. Positioned above Granite 4.0 350M in the Nano group, IBM also invites community fine-tuning for more languages.

1.6B params · dense131K contextApache 2.035.6K downloads58 likes8.1K quant downloads
S

Runs great on a 24 GB GPU

Runs from8 GB Q4_K_M · 64KBest on24 GB Q4_K_M · ~452 tok/s · 128KMacs from8 GB Q4_K_M · ~49 tok/s

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Granite 4.0 1B

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Recommended

Ollama
one command · needs Ollama installed
$ollama run hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_K_M

Q4_K_M fits fully on the RTX 4090 · 21.4 GB free · 16K context

Download: granite-4.0-1b-Q4_K_M.gguf1.02 GBibm-granite

install, OS, other quants

Ways to runOther ways

each opens step-by-step

Run in cloud

Your own private copy · Q4_K_M · 24 GB GPU

  • The easiest way to try this model
  • Running in a few minutes, easy setup
  • Chat right here or connect any app

$5+ bonus for new users

How Much VRAM Does Granite 4.0 1B Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KOn RTX 4090 (reference)
1.2 GB3.7 GB2-bit · K-quant — 2-bit quantization with K-quant improvementsFits: 128K context · ~565 tok/s
1.2 GB3.7 GB3-bit small — 3-bit small quantizationFits: 128K context · ~555 tok/s
1.3 GB3.8 GB3-bit medium — 3-bit medium quantizationFits: 128K context · ~520 tok/s
1.3 GB3.8 GB4-bit · legacy — 4-bit legacy quantizationFits: 128K context · ~512 tok/s
1.4 GB4 GB4-bit medium — 4-bit medium quantization — most popular sweet spotFits: 128K context · ~452 tok/s
1.6 GB4.2 GB5-bit medium — 5-bit medium quantization — good quality/size tradeoffFits: 128K context · ~402 tok/s
1.8 GB4.3 GB6-bit, very good — 6-bit quantization, very good qualityFits: 128K context · ~362 tok/s
2.1 GB4.6 GB8-bit, near-lossless — 8-bit quantization, near-losslessFits: 128K context · ~312 tok/s

Selected Q4_K_M

4-bit medium quantization

2.6GBat 16K context

Weights
1.0 GB
KV cache
1.3 GB
Overhead
0.3 GB

Fits RTX 4090 (reference) · 21.4 GB free · ~452 tok/s

≥50 tok/s needs ≥111 GB/s · RTX 5090 ~803, M5 Ultra ~593

Which Hardware Can Run Granite 4.0 1B?

Q4_K_M · 1.4 GB

Granite 4.0 1B (Q4_K_M) requires 1.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 2+ GB is recommended. Using the full 131K context window can add up to 10.6 GB, bringing total usage to 12.0 GB. 114 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 185 devices with unified memory can run Granite 4.0 1B, including Mac Studio M5 Ultra (256 GB), Mac Studio M5 Ultra (512 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M
8 GB · RTX 4060Q4_K_M · 2.6 GB · ~122 tok/s · 64K8 GB tier →
12 GB · RTX 3060 12GBQ4_K_M · 2.6 GB · ~161 tok/s · 64K12 GB tier →
16 GB · RTX 5060 Ti 16GBQ4_K_M · 2.6 GB · ~201 tok/s · 128K16 GB tier →
24 GB · RTX 4090referenceQ4_K_M · 2.6 GB · ~452 tok/s · 128K24 GB tier →
Mac 32 GB · Mac Mini M4Q4_K_M · 2.6 GB · ~58 tok/s · 128KMacs →
Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 2.6 GB · ~264 tok/s · 128KMacs →

How good is Granite 4.0 1B?

llmrun Score blends public benchmark boards into one number; 100 = the average open model.

Not enough data yet for an llmrun Score.

A Score needs results on enough independent boards. How the Composite Score is built

Benchmarks

Here are all 2 boards that have results for this model, sorted by their weight in the llmrun Score; badges show who ran the test.

  • GPQA Diamond

    Epoch AI · Oct 2025Independent

    24.0%

    #93/100 open

    Independent
  • AIME 2024/2025

    Epoch AI · Oct 2025Independent

    0.8%

    #83/88 open

    Independent

Alternatives to Granite 4.0 1B

Granite family · 28 models →

Similar models

6
ModelScoreVRAMtok/sWhy it is similar
LFM3 1.7BLiquid AI—2.1 GB~455same size
MiniCPM V 4.7 1BOpenBMB—1.6 GB~512same size
LFM2 VL 1.6BLiquid AI—1.5 GB~496same size
Gemma 4 31B IT DFlashz-lab—1.4 GB~524same size
Qwen3.6 27B DFlashz-lab—1.8 GB~471same size
Qwen Base Invoicev1.01 1.5BLaaP-ai—1.7 GB~512same size
Fine-tunes & variants(1)

Specifications

Publisher
IBM
Family
Granite
Parameters
1.6B
Architecture
GraniteMoeHybridForCausalLM
Context Length
131,072 tokens
Vocabulary Size
100,352
Release Date
2025-10-07
License
Apache 2.0

Frequently Asked Questions

How much VRAM does Granite 4.0 1B need?

Granite 4.0 1B requires 1.4 GB of VRAM at Q4_K_M, or 3.7 GB at BF16. Full 131K context adds up to 10.6 GB (12.0 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 1.6B × 4.8 bits ÷ 8 = 1 GB

KV Cache + Overhead ≈ 0.5 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 11 GB (at full 131K context)

VRAM usage by quantization

1.4 GB
12.0 GB

Learn more about VRAM estimation →

What's the best quantization for Granite 4.0 1B?

For Granite 4.0 1B, Q4_K_M (1.4 GB) offers the best balance of quality and VRAM usage. Q5_0 (1.5 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 0.9 GB.

VRAM requirement by quantization

IQ2_XXS
0.9 GB
Q3_K_M
1.3 GB
Q4_1
1.4 GB
Q4_K_M ★
1.4 GB
Q5_1
1.6 GB
BF16
3.7 GB

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Granite 4.0 1B on a Mac?

Yes — MacBook Air 13" M3 (8 GB) and 158 other Macs can run Granite 4.0 1B. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 1.4 GB of usable unified memory (RAM minus macOS overhead).

Can I run Granite 4.0 1B locally?

Yes — Granite 4.0 1B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Granite 4.0 1B?

At Q4_K_M, Granite 4.0 1B can reach ~3310 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~452 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 1.45 × 0.65 = ~3586 tok/s

Estimated speed at Q4_K_M (1.4 GB)

~3586 tok/s
~452 tok/s
~3586 tok/s
~3310 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Granite 4.0 1B?

At Q4_K_M, the download is about 0.98 GB. The full-precision BF16 version is 3.26 GB. The smallest option (IQ2_XXS) is 0.45 GB.

Which GPUs can run Granite 4.0 1B?

114 consumer GPUs can run Granite 4.0 1B at Q4_K_M. Top options include AMD Radeon RX 5600 XT, AMD Radeon RX 5700 XT, AMD Radeon RX 580 8GB. 114 GPUs have plenty of headroom for comfortable inference.

Which devices can run Granite 4.0 1B?

185 devices with unified memory can run Granite 4.0 1B at Q4_K_M (1.4 GB), including AMD Ryzen 7 8845HS (Hawk Point) Laptop, AMD Ryzen AI 7 350 (Krackan Point) Laptop, AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.