model release

Google Launches Gemini Nano 4 and Gemini Intelligence on Samsung's Galaxy Z Fold 8, Flip 8

TL;DR

Samsung's Galaxy Z Fold 8, Fold 8 Ultra, and Flip 8 are the first devices to ship with Google's Gemini Nano 4 on-device model and the new Gemini Intelligence feature tier. The launch comes with strict hardware requirements including 12GB+ RAM and qualified system-on-chips.

3 min read
0

Google's Gemini Nano 4 and Gemini Intelligence have launched for the first time on Samsung's Galaxy Z Fold 8, Fold 8 Ultra, and Flip 8, according to a Google developer blog post published Wednesday. The foldables mark the debut hardware for both the updated on-device model and Google's premium AI feature tier for Android.

Gemini Nano 4 is built on Gemma 4, Google's open model family, and supports more than 140 languages, according to Google. The company also claims the model offers improved multimodal understanding over prior Nano versions, though no specific benchmark scores were disclosed.

Hardware requirements

The Fold 8 and Fold 8 Ultra run on Qualcomm's Snapdragon 8 Elite Gen 5 for Galaxy, paired with either 12GB or 16GB of RAM. The Flip 8 instead uses Samsung's own Exynos 2600 chipset with 12GB of RAM.

To support Gemini Intelligence, Google has rewritten its Android developer requirements. The previous "Gemini Nano v3 or higher" spec on android.com has been replaced with a simpler "Nano models on device AI" requirement. The "Qualified SOC" criterion also dropped its previous "flagship" qualifier.

The full list of Gemini Intelligence requirements now includes: on-device Nano model support, the latest media performance class, 12GB or more of RAM, a qualified system-on-chip, meeting service-level objectives in the field through 2026 (with enforcement tightening in 2027), and passing Google's test suite at launch on Android 17 or higher.

What's shipping now vs. later

Google is treating expanded task automation — now covering more than 40 apps — as the first Gemini Intelligence feature to reach these devices. The company had said in May, when Gemini Intelligence was first announced, that additional capabilities would roll out in waves rather than all at once.

Still pending: Create My Widget, a Gboard feature called Rambler, and an Autofill with Google capability that draws on Gemini Personal Intelligence. Google has not given a timeline for when these will reach the Fold 8, Flip 8, or other qualifying devices.

All three Samsung devices carry a "with Gemini Intelligence" badge, which Google says is reserved for "premium hardware" meeting the full spec list above. The upcoming Pixel 11 series is expected to carry the same badge, based on Google's own qualification criteria, though Google has not confirmed this directly.

What this means

Gemini Intelligence functions as a hardware gatekeeper, not just a software feature — Google's spec list ensures only devices with sufficient RAM, approved chipsets, and current media performance classes qualify for the badge. That creates a two-tier Android AI experience: baseline Gemini Nano features for most phones, and a more capable feature set locked to premium hardware from 2026 onward.

The staggered feature rollout, with core capabilities like Create My Widget and Autofill with Google still unavailable at launch, suggests Google is prioritizing hardware certification and badge distribution over feature completeness. Whether the "with Gemini Intelligence" designation becomes a meaningful purchase driver — similar to how 5G or high refresh-rate displays became checkbox features — remains untested until Pixel 11 devices and other Android OEMs adopt the same requirements.

Related Articles

model release

iFLYTEK releases Spark X2.6, a 293B-A30B MoE model trained on domestic compute

At its 1024 developer festival, iFLYTEK released Spark X2.6, a 293B-A30B mixture-of-experts model trained on domestic compute. The company also open-sourced the on-device Spark X2.6-4B and debuted the SparkAudio 1.0 speech model.

model release

Recursive releases Flint, a 40B-parameter open coding model that runs on laptops under Apache 2.0

Recursive released Flint, a free, open-weight mixture-of-experts coding model with 40 billion total and 2.5 billion active parameters, under Apache 2.0. The company claims its Eureka Machine automated research system contributed to nearly every stage of development, and that Flint runs at about 131 tokens per second on an M4 Pro MacBook Pro.

model release

Hugging Face releases Carbon-X, a 14B-parameter open model that answers natural-language questions about DNA

Hugging Face released Carbon-X, a 14-billion-parameter open model that takes DNA sequences and English questions and returns conversational answers. It was trained on 300 billion tokens and is designed to run locally. Its developers say it has not yet been validated with wet lab experiments.

model release

Ant Digital Technologies releases LingDT-Decision-1.0 in 4B and 27B sizes for direct structured decisions

Ant Digital Technologies has released LingDT-Decision-1.0, a model that makes structured choices directly instead of generating text. It comes in 4B and 27B sizes, and the company says it cuts token use by more than 90%.

Comments

Loading...