4 verified stories covering Technology, product updates and industry developments.
Quantization and distillation are the two dominant technical approaches for shrinking and accelerating large models, enabling deployment on resource-constrained hardware.
Major AI models now offer million-token context windows, but accuracy and cost remain significant challenges.
MiniMax's 01 series introduces Lightning Attention, a linear attention variant that cuts computational complexity from O(n²) to near O(n), delivering significant speed gains in long-context scenarios while maintaining competitive accuracy.
DeepSeek has achieved legendary cost efficiency in training large AI models, with V3 costing about $5.5 million and R1 GPU usage at roughly $294,000, compared to U.S. competitors spending $100 million to $1 billion for similar-scale models.