What are the hardware requirements for GigaChat 20B A3B Instruct?

GigaChat 20B A3B Instruct requires approximately 82.5 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed GigaChat 20B A3B Instruct and how large is it?

GigaChat 20B A3B Instruct is developed by ai-sage, a model with 20b 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 GigaChat 20B A3B Instruct?

GigaChat 20B A3B Instruct is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

GigaChat 20B A3B Instruct by ai-sage — VRAM 82.5GB, 128K context

Name: GigaChat 20B A3B Instruct
Author: ai-sage

GigaChat 20B A3B Instruct is an open-source language model by ai-sage. Features: 20b LLM, VRAM: 82.5GB, Context: 128K, License: mit, Instruction-Based, LLM Explorer Score: 0.17.

Arxiv:2506.09440 Base model:ai-sage/gigachat-20... Base model:finetune:ai-sage/gi... Conversational Custom code Deepseek En Instruct Region:us Ru Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/ai-sage/GigaChat-20B-A3B-instruct

GigaChat 20B A3B Instruct Parameters and Internals

LLM Name	GigaChat 20B A3B Instruct
Repository 🤗	https://huggingface.co/ai-sage/GigaChat-20B-A3B-instruct
Base Model(s)	ai-sage/GigaChat-20B-A3B-base ai-sage/GigaChat-20B-A3B-base
Model Size	20b
Required VRAM	82.5 GB
Updated	2026-08-04
Maintainer	ai-sage
Model Type	deepseek
Instruction-Based	Yes
Model Files	5.0 GB: 1-of-17 5.0 GB: 2-of-17 5.0 GB: 3-of-17 5.0 GB: 4-of-17 5.0 GB: 5-of-17 5.0 GB: 6-of-17 5.0 GB: 7-of-17 5.0 GB: 8-of-17 5.0 GB: 9-of-17 5.0 GB: 10-of-17 5.0 GB: 11-of-17 5.0 GB: 12-of-17 5.0 GB: 13-of-17 5.0 GB: 14-of-17 5.0 GB: 15-of-17 5.0 GB: 16-of-17 2.5 GB: 17-of-17
Supported Languages	ru en
Model Architecture	DeepseekForCausalLM
License	mit
Context Length	131072
Model Max Length	131072
Transformers Version	4.47.0
Tokenizer Class	PreTrainedTokenizerFast
Vocabulary Size	128256
Torch Data Type	float32

Rank the GigaChat 20B A3B Instruct 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 GigaChat 20B A3B Instruct

Best Alternatives	Context / RAM	Downloads	Likes
GigaChat 20B A3B Instruct Bf16	128K / 41.2 GB	48	16