Philip Kiely(@philipkiely):It's the summer of Flash models. First GLM, now DeepSeek, these midsize models push the Pareto Frontier with remarkable intelligence at low prices. And, they outperform full-size last-gen models w/ new efficient architectures. Excited to see these recipes scale to 2T+ params.

Philip Kiely

@philipkiely

Author of Inference Engineering | Building @baseten | Not an LLM (yet)

参加 December 2018

970 フォロー中 13.8K ファン

Philip Kiely@philipkiely

2026.09.10 19:27

It's the summer of Flash models. First GLM, now DeepSeek, these midsize models push the Pareto Frontier with remarkable intelligence at low prices. And, they outperform full-size last-gen models w/ new efficient architectures. Excited to see these recipes scale to 2T+ params.

DeepSeek@deepseek_ai

2026.09.10 06:10

🚀 Introducing DeepSeek-V4.1-Flash: smarter, faster, more efficient. 🔹 Introducing the smallest model in our new architecture family, with native visual understanding. 🔹 Designed for greater capability, faster inference, higher throughput, and scaling to larger models. 1/6