Skip to content
worth noting Hardware

Qualcomm introduced Snapdragon 8 Elite Gen 6 and Extreme Gen 6 processors for running AI models locally on phones

only one source so far

Qualcomm announced the Snapdragon 8 Elite Gen 6 and Extreme Gen 6 mobile processors focused on local AI. According to the company, the Extreme version can run a 30-billion-parameter MoE model directly on the phone, while the new sensing hub runs models with up to 200 million parameters for voice agents.

Qualcomm announced two new flagship smartphone processors at Snapdragon Summit – Snapdragon 8 Elite Gen 6 and Snapdragon 8 Elite Extreme Gen 6. According to the company, both chips are intended to enable more advanced AI features that run directly on the device. The new sensing hub can run small models with up to 200 million parameters locally – allowing the phone to offer a local voice recorder with speaker identification, build a memory of usage patterns to improve suggestions for task automation, and run a complete voice agent (with both voice input and output) without the need for the cloud.

Snapdragon 8 Elite Gen 6 includes a new accelerator for running models more efficiently. According to the company, the more powerful Extreme variant can run a mixture-of-experts (MoE) model with 30 billion parameters locally, with only some of the parameters activated for a specific task. For comparison, the company notes that Apple introduced a MoE model with 20 billion parameters at the WWDC developer conference in June as the most advanced model in its third generation of foundation models.

The new chips also bring camera features – pixel-level capture control, improved stabilization, and motion recognition. The Extreme version supports video recording at up to 8K at 60 fps and 4K at 240 fps for slow-motion footage, and supports the new Advanced Professional Video (APV) codec. Both chips can also use AI to enhance voice and suppress noise, including the new voice bubble technology from Qualcomm for isolating ambient noise during calls.

Following the announcement, Motorola introduced the Motorola Signature 27 phone, powered by Snapdragon 8 Elite Extreme Gen 6, which is expected to be available this year. According to the source, Qualcomm is working on more than 40 devices with AI and, along with other voices in the industry (Nothing co-founder Carl Pei, Apple chief John Ternus), expects phones to remain the primary device for using AI.

What changed

Why it matters

The new chips push the limits of which AI models can run directly on a phone without the cloud – reducing reliance on a network connection and limiting the amount of data sent to servers. For mobile app developers, this opens up the possibility of deploying larger local models (voice agents, personalization) directly on the hardware, changing the requirements for designing AI features in mobile products.

Two audiences, two different impacts

What this means

01

For individuals

Future phones with Snapdragon 8 Elite Gen 6/Extreme Gen 6 (e.g. Motorola Signature 27) will offer voice AI agents and personalization that work locally without the need to send data to the cloud.

What to do When choosing a new phone, consider models with Snapdragon 8 Elite Gen 6 or Extreme Gen 6 if interested in local voice AI features.
More practical updates →
02

For a business

Mobile app developers will be able to deploy larger local AI models (up to 30B MoE on the Extreme variant) directly on the device, reducing reliance on cloud inference and changing the options for designing AI features in mobile products.

Development
What to decide Monitor the availability of developer tools for the new Snapdragon chips when planning mobile AI features that run locally on the device.
More business impacts →
AI akcelerátor místní AI modely mobilní procesory MoE Qualcomm Snapdragon

Check the original

Event sources

only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.

1
TechCrunch AI independent context · first detected Qualcomm launches two new smartphone chips with emphasis on AI