Hugging Face has released version 5.14.0 of its Transformers library, introducing new model additions and performance enhancements. The update includes the Inkling multimodal model from Thinking Machines, which supports text, image, and audio inputs and generates text outputs. This release also features improvements to kernel operations, generation capabilities like Multi-Token Prediction, and cache management, alongside performance fixes for Flash Attention and MoE decoding. AI
IMPACT Enhances developer tools for building AI applications with new multimodal capabilities and performance optimizations.
RANK_REASON This is a software library release with new model additions and performance improvements, not a frontier model release from a major lab.
Read on Transformers — Releases →
- FlashAttention
- GPTBigCode
- GPTNeoX
- Hugging Face
- Inkling
- Qwen3-VL
- Thinking Machines
- TIPSv2
- Transformers
- vLLM
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
