Mercury 3 is Inception's most capable diffusion LLM, over 40 times the size of Mercury 1 and generating more than 1,000 tokens per second, with quality on par with frontier speed-optimized models across a 1M-token context.
Mercury 3 is Inception's most capable diffusion LLM, over 40 times the size of Mercury 1 and generating more than 1,000 tokens per second, with quality on par with frontier speed-optimized models across a 1M-token context.