AI Tools & Products
Oct 19, 2026
Launch of Mercury 3: A New High-Performance Diffusion-Based Language Model
Oct 19, 2026
AI Summary
Mercury 3 has been introduced as a diffusion-based language model that significantly outperforms its predecessors in speed and intelligence. It is designed for various applications, including real-time tutoring and customer support, and is available at a promotional price for early users.
- Mercury 3 is the latest diffusion-based language model, outperforming GPT-6 Luna and Claude Haiku 5.5 across multiple benchmarks.
- It operates at speeds of 1,068 tokens per second and has a context capacity of 1 million tokens, an increase from 260,000 in Mercury 2.5.
- The model is 40 times larger than its predecessor, Mercury 1, and demonstrates a threefold increase in intelligence based on standard benchmarks.
- Early-access partners are utilizing Mercury 3 for applications such as voice and chat support, real-time tutoring, and data engineering.
- The pricing for Mercury 3 is set at $0.20 per million input tokens and $0.70 per million output tokens, with a launch discount of 50%.
- Mercury 3 has shown to complete tasks more efficiently than GPT-6.1-Sol in live navigation tasks, maintaining low costs per token.
- The model is compatible with existing systems and offers enterprise solutions with dedicated capacity and compliance controls.
- Additional products in the Mercury lineup include Mercury Voice for low-latency voice applications and Mercury Decide for high-volume classification tasks.
mercury 3diffusionscalable aiinception labsai tools