All Articles  (X)

MAX 25.2 is turning heads — and for good reason. This powerful update delivers industry-leading performance for large language models on NVIDIA GPUs, all without CUDA. MAX 25.2 builds on the momentum of 25.1 and introduces major upgrades to help you build GenAI systems that are faster, leaner, and easier to scale.

In this post, we’ll break down how Python eDSLs work, their strengths and weaknesses, and take a close look at Triton.

We’re excited to announce MAX 25.2, a major update that unlocks industry-leading performance on the largest language models–built from the ground up without CUDA.

We recently introduced MAX 25.1, a major leap forward in AI development. This release enhances agentic and LLM workflows, introduces MAX Builds as a central hub for GenAI models and application recipes, and debuts a new GPU programming interface. Developers can now take advantage of GPU-accelerated embeddings, OpenAI-compatible function calling, structured output generation, and high-performance LLM optimizations like paged attention and prefix caching for improved efficiency.

Answering the question of whether CUDA is “good” is much trickier than it sounds.

If we as an ecosystem hope to make progress, we need to understand how the CUDA software empire became so dominant.