Models & Labs
Kimi K3: Largest Open-Weight AI Model Released
Together AI Bloghigh confidence
Why it matters
- →Kimi K3 sets a new benchmark as the largest open-weight AI model, expanding possibilities for open-source AI development.
- →Its advanced architecture supports complex tasks, making it a competitive alternative to proprietary models.
- →The model's release enhances accessibility to cutting-edge AI capabilities for developers and researchers.
Moonshot AI has released Kimi K3, a 2.8-trillion-parameter model, making it the largest open-weight AI model to date. Designed for advanced tasks such as long-horizon coding and deep reasoning, Kimi K3 competes with leading proprietary models like GPT 5.6 Sol. It features new architectural innovations, including Kimi Delta Attention and Attention Residuals, which improve its ability to process long sequences and complex tasks. This release represents a significant advancement in open-source AI, providing developers with a powerful new tool for sophisticated AI applications.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
More in Models & Labs
Models & Labsother
llama.cpp b12340 adds CUDA 13 and ROCm 10.0
This release quietly closes the hardware gap for local inference by adding default support for CUDA 13 and ROCm 10.0 across Linux and Windows. Users with newer NVIDIA GPUs or AMD accelerators no longer need to compile from source to get performance, while Apple Silicon builds now ship KleidiAI disabled by default, shifting the focus back to standard ARM optimizations. The inclusion of Snapdragon support for Linux and Android marks a significant step toward on-device AI beyond the x86/ARM desktop ecosystem. It is less about new features and more about ensuring llama.cpp remains the universal runtime for every available accelerator.
llama.cpp Releases
Models & Labscoding
llama.cpp b12342 adds few-row MMQ for speculative decoding
Speculative decoding just got a significant speed boost on NVIDIA GPUs. This release introduces a specialized kernel for 'few-row' matrix multiplications, targeting the small batch sizes typical of verification passes in speculative decoding workflows. By optimizing these specific operations for quantized formats like Q8_0 and IQ4 variants, llama.cpp reduces latency where it matters most during token generation. The change is CUDA-only for now but leverages hardware capabilities from Ampere onwards, making local inference faster without requiring new model architectures or complex configuration changes.
