AI Intel

Ollama v0.40.0-rc0 update now defaults to running MLX-supported model architectures on Apple Silicon devices.

Unsloth AI announced that GLM-5.3-Flash can now be run locally with Unsloth GGUF, enabling 3-bit operation on 128GB RAM and rivaling Claude Opus 4.8…

Kimi_Moonshot's Kimi K3 open-weight model is now generally available and rolling out in GitHub Copilot, demonstrating frontier-level agentic coding a…

Mirage unveiled Tesseract, a video creative suite rebuilt for GPT-6 Astra and Sol, allowing AI agents direct access to its creative engine for After…

ChatGPT Work has introduced a new Data agent, enabling users to transform company data into answers, interactive dashboards, and actions simply by as…

DeepSeek-V4-Pro API is offering a 75% discount until May 5th, 2026, and now supports a 1M context through integration updates.

vLLM has released v0.30.1rc0, incorporating MI355 dense NVFP4 and MoRI kernel mirrors.

llama.cpp release b11176 addresses a bug in tensor split for fused QKV when K/V head sizes are uneven.

Former Intel CEO and other industry experts criticize the limitations of HBM (High Bandwidth Memory), signaling the potential emergence of high bandw…

Go is experimenting with implementing platform-independent SIMD (Single Instruction, Multiple Data) to boost performance.