AMD GenAI

Featured Releases

Instella-Math

First Math LLM with Long CoT RL on AMD GPUs

Learn more →

Instella-T2I

Pushing the Limits of 1D Discrete Latent Space Image Generation

Learn more →

Instella-Long

A Fully Open Language Model with Long-Context

Learn more →

AMD GenAI advances the frontier of AI through open innovation and collaboration. We build fully open-source foundation models optimized for AMD GPUs such as MI250 and MI300, and release everything -- code, models, and training recipes -- to the public.

Our mission is to empower AI researchers and engineers to train and develop cutting-edge models efficiently on AMD systems.

AMD GenAI is part of AMD. To learn more about AMD, visit AMD.

Recent Blogs

GEAK-Triton v2 Family of AI Agents: Kernel Optimization for AMD Instinct GPUs

December 23, 2025

Announcing GEAK-OptimAgentv2 and GEAK-OpenEvolve AI agents for generating Triton kernels optimized for AMD Instinct GPUs, featuring multi-offspring evolution, hardware-aware feedback, and quality-diversity search for up to 3.42× speedups.

GEAK-HIP: Expanding GEAK for HIP Code Optimization

December 23, 2025

Extending the GEAK family to native HIP code generation and optimization, enabling AI-driven kernel development for AMD Instinct GPUs with similar techniques applied to HIP kernels.

SAND-Math: Building a 32B Reasoning Model with Synthetic Data

December 06, 2025

Building a state-of-the-art 32B reasoning model on AMD Instinct™ MI325X GPUs using only synthetic data. SAND-MathScience-DeepSeek-Qwen32B achieves 78.33% on AIME25, surpassing Qwen3-32B, with fully open-sourced model weights, datasets, and training code.

Read All Blogs

Recent Publications

Publication

ICLR 2026

ImageDoctor: Diagnosing Text-to-Image Generation via Grounded Image Reasoning

ImageDoctor diagnoses text-to-image generation through grounded image reasoning, enabling systematic evaluation and debugging of T2I models by identifying misalignments between prompts and generated images.

Read more

Publication

ICLR 2026

XModBench: Benchmarking Cross-Modal Capabilities and Consistency in Omni-Language Models

XModBench is a comprehensive benchmark for evaluating cross-modal capabilities and consistency in omni-language models, assessing how well models handle information across different modalities.

Read more

Publication

ICLR 2026

Latent Visual Reasoning

Latent Visual Reasoning enables models to perform complex visual reasoning in latent space, improving efficiency and accuracy on multi-step visual reasoning tasks.

Read more

Read All Publications