A new model from Tencent, Hy4 flash, has been released and is now accessible via Selectel's AI router. This model, part of the Hy4 family, boasts 224 billion parameters and features an Engram memory capable of handling a million tokens, accepting both text and image inputs. While Tencent has not yet released the model weights, it can be accessed through an OpenAI-compatible API provided by Selectel, with payment based on token usage. AI
IMPACT This release offers a large context window and multimodal capabilities, potentially impacting how developers build applications requiring extensive memory.
RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=2 ai=1.0]
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →