Summary: Together AI adds new video and image models

Why It Matters: This shows Together AI is growing its offerings beyond just text models. It means they want to serve more kinds of AI users.

What Changed:

• The main title changed from "Together AI – The AI Acceleration Cloud - Fast Inference, Fine-Tuning & Training" to "Fast, Reliable AI Inference at Scale | Together AI".
• The cookie consent message was updated to be more detailed.
• New blog posts were added to the top section, including "ATLAS: runtime-learning accelerators," "Together Instant Clusters," "Batch Inference API," and "Fine-Tuning Platform Upgrades."
• A new "Startup Accelerator" solution was added.
• A new "Events" section was added under Resources.
• The main hero section changed from "The AI Acceleration Cloud" to "Serverless inference" with a focus on running inference with an API call.
• The list of featured models on the homepage was updated.
• Many new image and video generation models were added to the model library, including "Gemini Flash Image 2.5," "Sora 2 Pro," "Google Imagen 4.0 Ultra," "ByteDance Seedream 3.0," "HiDream-I1-Dev," "Juggernaut Lightning Flux," "Qwen Image," "Ideogram 3.0," "Stable Diffusion 3," "ByteDance Seedance 1.0 Pro," "MiniMax Hailuo 02," "Kling 2.1 Master," "Google Veo 3.0," and "Vidu 2.0."
• Some existing models were updated or removed from the visible list.