Samsung and Nota AI expand partnership to optimize AI on Exynos 2600 with up to 90% reduction in model size

Samsung Electronics has entered into an agreement with South Korean company Nota AI to integrate artificial intelligence model optimization technology into the Exynos 2600 processor. Essa platform, called NetsPresso, allows you to reduce the size of large language models by up to 90% without loss of accuracy. The chip should equip part of the smartphones in the Galaxy S26 line, scheduled for launch in 2026.
The collaboration extends the partnership initiated in previous versions of the Exynos, such as the 2500 and 2400. With optimization, the Exynos 2600 makes it possible to execute generative AI functions directly on the device, without dependence on an internet connection. Isso improves speed, privacy and energy efficiency in complex tasks.
The Exynos 2600, manufactured on a 2 nm process with gate-all-around (GAA) technology, records significant gains in AI performance, with an increase of up to 113% compared to its predecessor.
NetsPresso platform in detail
NetsPresso compresses AI models while maintaining high accuracy across different hardware.
This drastic size reduction makes it easier to operate larger models locally on smartphones.
Developers gain automation in the optimization pipeline to quickly implement new features.

Continuity of collaboration
The partnership between Samsung and Nota AI began with the Exynos 2400 and continued with the 2500, used in foldable devices.
The current agreement includes joint development of the next version of the Exynos AI Studio model optimization tool for the chip.
Myungsu Chae, CEO of Nota AI, highlighted that integration with Samsung hardware generates concrete value in on-device AI.
Exynos 2600 Key Specifications
- 10-core CPU with updated Arm architecture, offering up to 39% more performance.
- GPU Xclipse 960 based on AMD’s RDNA 4, with significant gains in graphics.
- Improved NPU for AI tasks, supporting advanced instructions like SME2.
- Heat Path Block cooling system for thermal control in intense use.
Practical applications on the Galaxy S26
The processor equips at least part of the Galaxy S26 family, with possible regional use ranging between Exynos and Snapdragon.
Users access offline generative AI for image editing, real-time translation, and advanced assistants.
- Local execution guarantees more agile responses.
- Reduces battery consumption compared to cloud processing.
- Increases privacy by avoiding sending data to external servers.
- Allows resources in unconnected areas.
Benefits for developers
Automation in the Exynos AI Studio simplifies adaptation of recent models to the hardware.
This accelerates the creation of applications with embedded AI.
Optimization maintains efficiency even on large, demanding models.
Expectations for the launch
The Galaxy S26 line arrives between February and March 2026, according to current projections.
Tests on end devices reveal real-world performance from the combination of hardware and software optimization.
The technology positions the Exynos 2600 as a competitive option in mobile AI.






