Upstage Has Released Solar Mini 4 AI Model
The new model utilizes a mixture of experts design to improve efficiency for users.
Updated on Oct. 1, 2026 in Artificial Intelligence
Live Poll
Would you prioritize using smaller, more cost-efficient AI models for your business or personal tasks?
Upstage released the Solar Mini 4 large language model on October 1, 2026. The new tool features 35 billion parameters and a mixture of experts structure to optimize performance.
Why it matters
The model design aims to improve computational efficiency and reduce costs for developers. By using selective activation, it offers a high-performance option for those with limited hardware.
Solar Mini 4 features 35 billion total parameters, with only 3 billion activated during inference. The model utilizes quantization to operate on a single graphics processing unit.
The players
Upstage
Upstage is a technology company specializing in the development of advanced artificial intelligence models and large language solutions.
OpenRouter
OpenRouter is an API platform that aggregates access to various large language models for developers and researchers.
The details
The model uses a mixture of experts structure to selectively activate parameters, which helps in maintaining lower operational costs. Users can access the technology through the OpenRouter API platform.
Timeline
Upstage released Solar Mini 4 on October 1, 2026.
Discounted API pricing is available until October 10, 2026.
The Tech Race
This release follows the industry trend of utilizing the Mixture of Experts architecture to maintain model performance while reducing compute requirements. It signals a move away from massive dense models toward more efficient designs that favor accessibility.
Developers can now run high-performance models on a single graphics processing unit, significantly lowering the barrier to entry for local hardware usage. The current 70% discount provides a cost-effective window for early testing on the OpenRouter platform.
The takeaway
The move toward sparse, highly efficient models allows for better performance with fewer hardware constraints. Developers should leverage this window of reduced pricing to test how efficient architectures can fit into their existing workflows.
What happens next
A 70% discount on API prices for the model remains in effect through October 10, 2026.
Further reading
For more information on the evolving landscape of model architecture, visit our Artificial Intelligence section.
Source note: This article includes information reported by 조선일보.
Live Poll
Would you prioritize using smaller, more cost-efficient AI models for your business or personal tasks?
