MiniMax has released M3.1 Flash, a new agentic model that is three times faster and 70% cheaper for tasks like coding and cybersecurity, with plans to release model weights later. Inception Labs introduced Mercury 3, a diffusion-based language model that reportedly surpasses GPT-6 Luna and Claude Haiku 5.5 in benchmarks while achieving over 1,000 tokens per second. Additionally, Exa 3.0 has been launched, excelling in coding, finance, and research, and includes a new search product called Exa Lite. AI
IMPACT New models from MiniMax, Inception Labs, and Exa offer significant speed and cost improvements, potentially accelerating adoption of advanced AI capabilities.
RANK_REASON Multiple frontier labs (MiniMax, Inception Labs, Exa) announced new model releases with performance and cost improvements. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- ChatGPT
- Clairic
- Claude
- Claude Haiku 5.5
- Exa 3.0
- Gemini
- GPT-6 Luna
- Inception Labs
- International Monetary Fund
- M3.1 Flash
- Mercury III
- Meta*
- Minimax
- MiniMax API
- MiniMax Code
- Muse
- Prometheus
- PyTorch
- System One
- TorchNeuron
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →