arXiv:2607.02770 — Paper

Specification

No structured attributes yet.

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

0

Source tiers

—

Freshest observation

—

Conflicts

None

gemma-4-12B-it-qat-w4a16-ct GoogleActiveOpen weights
params 13.3Breleased 5 Jun 2026license apache-2.0
53
gemma-4-12B-it GoogleActiveOpen weights
params 12Breleased 23 May 2026license apache-2.0
53
gemma-4-E2B-it GoogleActiveOpen weights
params 5.12Breleased 2 Mar 2026license apache-2.0
53
gemma-4-E4B-it GoogleActiveOpen weights
params 8Breleased 2 Mar 2026license apache-2.0
53
Gemma 4 31B GoogleActiveOpen weights
Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...
params 31.3Bcontext 262.1Kreleased 11 Mar 2026license apache-2.0
65
Gemma 4 26B A4B GoogleActiveOpen weights
Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...
params 25.8B (4B active)context 262.1Kreleased 11 Mar 2026license apache-2.0
65

Describes: gemma-4-12B-it-qat-w4a16-ct, gemma-4-12B-it, gemma-4-E2B-it, gemma-4-E4B-it, Gemma 4 31B, Gemma 4 26B A4B

As of

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

Claim history

No claims recorded yet

Claims appear when a source states a fact; every later change is kept as a new claim.

11 Sept 2026
New paperPaperarXiv:2607.02770
New paper: arXiv:2607.02770
huggingface
huggingface.co

Full timeline →

Source documents
Source	Document	Type	Tier	Last observed	Snapshots
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/google/gemma-4-12B-it-qat-w4a16-ct	model_page	T2· Quality secondary	4 h ago	2
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/google/gemma-4-12B-it	model_page	T2· Quality secondary	4 h ago	1
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/google/gemma-4-E2B-it	model_page	T2· Quality secondary	4 h ago	1
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/google/gemma-4-E4B-it	model_page	T2· Quality secondary	4 h ago	1
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/google/gemma-4-31B-it	model_page	T2· Quality secondary	4 h ago	1
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/google/gemma-4-26B-A4B-it	model_page	T2· Quality secondary	4 h ago	1

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