Specification

Release date: 15 Apr 2025
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Openness: open-weights
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
License: other
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Architecture: Gemma3ForConditionalGeneration
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Parameters: 4.97B
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Languages: multilingual
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Base model: OpenGVLab/InternVL3-1B-Instruct
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Quantization: 4bit
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Quantization format: mlx
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
File size: 3 GB
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Hugging Face repo: mlx-community/gemma-3-4b-it-qat-4bit
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Pipeline tag: image-text-to-text
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Model card: huggingface.co/mlx-community/gemma-3-4b-it-qat-4bit
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Downloads: 55,062
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 39 min agomedium
Likes: 9
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 39 min agomedium
Datasets: OpenGVLab/MMPR-v1.2
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Gated: No
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Is quantized: Yes
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Last modified: 2025-04-21T20:31:00+00:00
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Library name: transformers
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
License name: qwen
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 min agomedium
License url: https://huggingface.co/Qwen/Qwen2.5-72B-Instruct/blob/main/LICENSE
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Downloads all time: 3,286,987
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 39 min agomedium
Model type: gemma3
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Tags: transformers, safetensors, gemma3, image-text-to-text, internvl, custom_code, mlx, multilingual
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium
Weights dtype: BF16, U32
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium

Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →

Provenance

Attributed facts

26

Source tiers

T226

Freshest observation

4 min ago

Conflicts

None

Modalities

Modalities unavailable.

Capabilities

Tool calling
Unavailable
Structured output
Unavailable
Reasoning
Unavailable
Vision
Unavailable
Audio
Unavailable
Fine-tuning available
Unavailable

No capability flags have been observed from a source yet — we do not infer them.

Languages: multilingual
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 41 min agomedium

No benchmark results recorded

Results appear when a tier 1–3 source publishes them; we never copy scores without a source.

Current prices

No current prices recorded

Prices appear when a provider publishes a public pricing page we crawl.

Price history

Estimated

Memory need = bytes per parameter (4-bit ≈ 0.5 × 1.15 overhead, 8-bit 1.0, fp16 2.0) + a KV-cache allowance. Not a measurement.

Estimated hardware fit
Hardware	Quantization	Memory	Est. need	Fits
Apple M3 Ultra	4bit	—	3.4 GB	Yes
AMD Instinct MI325X	4bit	256 GB	3.4 GB	Yes
AMD Instinct MI300X	4bit	192 GB	3.4 GB	Yes
Apple M2 Ultra	4bit	—	3.4 GB	Yes
NVIDIA B200	4bit	180 GB	3.4 GB	Yes
NVIDIA H200	4bit	141 GB	3.4 GB	Yes
Apple M1 Ultra	4bit	—	3.4 GB	Yes
Apple M3 Max	4bit	—	3.4 GB	Yes
Apple M4 Max	4bit	—	3.4 GB	Yes
NVIDIA DGX Spark	4bit	128 GB	3.4 GB	Yes
Apple M2 Max	4bit	—	3.4 GB	Yes
NVIDIA A100 80GB	4bit	80 GB	3.4 GB	Yes
NVIDIA H100 SXM	4bit	80 GB	3.4 GB	Yes
Apple M1 Max	4bit	—	3.4 GB	Yes
Apple M4 Pro	4bit	—	3.4 GB	Yes
Apple M3 Pro	4bit	—	3.4 GB	Yes
Apple M1 Pro	4bit	—	3.4 GB	Yes
Apple M2 Pro	4bit	—	3.4 GB	Yes
Apple M4	4bit	—	3.4 GB	Yes
NVIDIA GeForce RTX 5090	4bit	32 GB	3.4 GB	Yes
Apple M2	4bit	—	3.4 GB	Yes
Apple M3	4bit	—	3.4 GB	Yes
NVIDIA GeForce RTX 3090	4bit	24 GB	3.4 GB	Yes
NVIDIA GeForce RTX 4090	4bit	24 GB	3.4 GB	Yes
Apple M1	4bit	—	3.4 GB	Yes

Ancestors 1

InternVL3-1B-InstructOpenGVLab

This model

mlx-community/gemma-3-4b-it-qat-4bit

4.97B params

Quantizations 0

None recorded.

Descendants 0

None recorded.

Papers 0

No papers linked yet.

Repositories 0

No repositories linked yet.

As of

Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.

Claim history · Model card

1 claims · 1 propertiesShow all properties

Model cardmodel_card_url1

Claim history for Model card
Value	Valid from → to	Status	Source	Confidence	Extractor
https://huggingface.co/mlx-community/gemma-3-4b-it-qat-4bit	11 Sept 2026 21:11 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

11 Sept 2026
New modelModelmlx-community/gemma-3-4b-it-qat-4bitMLX Community
New model: mlx-community/gemma-3-4b-it-qat-4bit (MLX Community)
huggingface
huggingface.co

Full timeline →

Source documents
Source	Document	Type	Tier	Last observed	Snapshots
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/mlx-community/gemma-3-4b-it-qat-4bit/raw/main/README.md	model_card	T2· Quality secondary	3 min ago	1
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/mlx-community/gemma-3-4b-it-qat-4bit	model_page	T2· Quality secondary	39 min ago	2
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/models?author=mlx-community&p=0&sort=downloads	listing	T2· Quality secondary	1 h ago	2

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.