What are the hardware requirements for Granite 3B Code Instruct 128K?

Granite 3B Code Instruct 128K requires approximately 7 GB of VRAM and supports a context window of 125K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Granite 3B Code Instruct 128K and how large is it?

Granite 3B Code Instruct 128K is developed by ibm-granite, a model with 3b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Granite 3B Code Instruct 128K?

Granite 3B Code Instruct 128K is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Granite 3B Code Instruct 128K by ibm-granite — VRAM 7GB, 125K context

Name: Granite 3B Code Instruct 128K
Author: ibm-granite

Granite 3B Code Instruct 128K is an open-source language model by ibm-granite. Features: 3b LLM, VRAM: 7GB, Context: 125K, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.14.

Arxiv:2405.04324 Code Conversational Dataset:bigcode/commitpackft Dataset:bigcode/self-oss-instr... Dataset:bugdaryan/sql-create-c... Dataset:garage-baind/open-plat... Dataset:glaive-function-callin... Dataset:glaiveai/glaive-code-a... Dataset:meta-math/metamathqa Dataset:nvidia/helpsteer Dataset:tiger-lab/mathinstruct Deploy:azure Granite Instruct Llama Model-index Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/ibm-granite/granite-3b-code-instruct-128k

Granite 3B Code Instruct 128K Parameters and Internals

Model Type

text-generation, code

Use Cases

Areas:

coding assistants, long-conext input processing

Limitations:

Performance may be limited with out-of-domain programming languages.

Considerations:

Provide few-shot examples to steer output.

Training Details

Data Sources:

CommitPackFT, BigCode-SC2-Instruct, MathInstruct, MetaMathQA, Glaive-Code-Assistant-v3, Glaive-Function-Calling-v2, NL2SQL11, HelpSteer, OpenPlatypus

Methodology:

Exposed to both short and long context data to enhance long-context capability without sacrificing code generation performance at short input context.

Hardware Used:

NVIDIA A100, NVIDIA H100

LLM Name	Granite 3B Code Instruct 128K
Repository 🤗	https://huggingface.co/ibm-granite/granite-3b-code-instruct-128k
Model Size	3b
Required VRAM	7 GB
Updated	2026-08-08
Maintainer	ibm-granite
Model Type	llama
Instruction-Based	Yes
Model Files	5.0 GB: 1-of-2 2.0 GB: 2-of-2
Model Architecture	LlamaForCausalLM
License	apache-2.0
Context Length	128000
Model Max Length	128000
Transformers Version	4.41.2
Tokenizer Class	GPT2Tokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	49152
Torch Data Type	bfloat16

Rank the Granite 3B Code Instruct 128K capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Granite 3B Code Instruct 128K

Best Alternatives	Context / RAM	Downloads	Likes
Llama 3.2 3B Instruct	128K / 6.5 GB	1829572	2368
Llama 3.2 3B Instruct	128K / 6.5 GB	186207	96
...3B TechWriter Instruct Unsloth	128K / 6.5 GB	4264	1
...lama 3.2 3B ORPO Advertisement	128K / 6.5 GB	189	2
...ld Llama32 3B Instruct Qa 1p25	128K / 6.5 GB	328	1
...old Llama32 3B Instruct Qa 2p5	128K / 6.5 GB	319	1
...ini Math345 GroupB Llama32 End	128K / 6.5 GB	755	0
...ama 3.2 3B TechWriter Instruct	128K / 6.4 GB	226	0
Llama 3.2 3B Instruct	128K / 6.5 GB	991	9
Alloma 3B Instruct	128K / 6.5 GB	55456	6

Note: green Score (e.g. "73.2") means that the model is better than ibm-granite/granite-3b-code-instruct-128k.