Tag
A tweet links to a Reddit thread about MiniMax issues, suggesting it deserves more attention than unrelated political posts.
Kijai's MiniMax-H3-experimental is a work-in-progress model featuring w4a8 quantization and int8_convrot VAE support for ComfyUI, released for testing purposes.
MiniMax announces open weights and open research, signaling a commitment to open innovation.
MiniMax raised $2B in funding; CEO commits to no salary until AGI and dedicates shares to open-source community.
MiniMax closed a $2B funding round; CEO IO pledges no salary until AGI, commits 4% equity to employees and 1% to open-source community.
MiniMax2.7 model released with 47 trillion parameters and 1200pp context length.
The next MiniMax AI model is expected to be released in about 10 days, but the tweeter anticipates it may be too large for their hardware.
MiniMax is rolling out its M3 model on the API, featuring a 1,000,000 token context window.
A developer tests the MiniMax M2.7 model via its API on three practical machine learning and coding workflows, evaluating its performance.
MiniMax launches a unified plan covering CLI, API, and Agent with shared credits across all models (M2.7, video, voice, music) and open-sources the entire stack including Teams and Mavis. Head of R&D hints at upcoming M3 model.
The MiniMax M2.7 model has been fine-tuned into an uncensored variant, 'ultra uncensored heretic', with very low refusal rates (4/100). Available in Safetensors and GGUF formats on HuggingFace.
Release of a mixed-bit quantized version of the MiniMax M2.7 model, optimized to 74 GB for efficient local inference on Apple Silicon devices.