Tracked topic
… AIM Catalog of AI Models Existing models Added in version 1.8 Cohere Labs Command A Reasoning 08 2025 DeepSeek DeepSeek R1 DeepSeek R1 0528 DeepSeek V3.1 DeepSeek V3.1 Terminus Qwen AI Qwen3 235B A22B Mistral AI Ministral 3 14B Instruct 2512 Qwen3 32B Mistral Large 3 675B Instruct 2512 Meta Llama 3… …
… Figure 1: Cost per million tokens vs. interactivity -- DeepSeek R1, FP8, no MTP. …
… Phase 2: Finals April 7 – May 15, 2026 Finalists will focus on end‑to‑end inference optimization of selected LLM workloads including DeepSeek‑R1 and Kimi 2K.5 aiming to achieve breakthrough performance on standardized benchmarks running on AMD Instinct MI355X GPUs. …
… Support for the latest Algorithms and Models With AMD ROCm 7, developers can deploy the latest open-source algorithms and models, including LLaMA 4, Gemma 3, and DeepSeek, from day one. …
… March 13, 2026 AMD Enterprise AI Suite Version 1.8 AMD Enterprise AI Suite Expands with DeepSeek and Mistral AI models, and adds Support for AMD Instinct™ MI350X and MI355X GPUs. …
… 4 They also offer up to 89% better chatbot performance on DeepSeek with AMD EPYC 9965 vs. …
… ROCm software delivers Day-0 support for AI platforms and models provided by leaders such as OpenAI, Meta, PyTorch, Hugging Face, xAI, DeepSeek, and more, thanks to strategic and deep collaborations with key partners. …
… Speculative mode: AMD PACE speculative decoding achieves 2.5–3.2x geometric mean speedup over vLLM autogressive mode across Llama 3.1-8B, Qwen2-7B, and DeepSeek-R1-7B at batch size 1, peaking at 4.45x on Llama with long output sequences. …