Inception Labs has released Mercury 3, a new model that reportedly matches or surpasses the performance of GPT-6 Luna and Haiku in certain general output tasks. Initial testing suggests Mercury 3 can process information at a rate of 1000 tokens per second, comparable to Luna and Haiku. Further evaluation is needed to confirm its capabilities, particularly in agentic frameworks. AI
IMPACT Potentially offers a new high-performance model for general tasks and agentic frameworks.
RANK_REASON New model release from a frontier lab. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →