New AI Models
Every notable new AI model from the major labs, newest first and curated from official announcements. Filter by provider and model type to see who shipped what, and when.
698 releasesPage 9 of 35
How to track new AI models
Four ways to read the board. Each one turns the stream of new AI models into a clear answer. Pick the view that fits your question.
Read the newest launches first
The board lists new AI models newest first, so the top row is always the latest launch from any lab. Every row links to the official announcement, so one click gets you to the source. Dates reflect the initial public release.
Follow one provider
Pick a lab in the Provider filter. OpenAI, Anthropic, Google, Meta, DeepSeek, Qwen and 20+ more each get their own view. Use it to read a lab's release rhythm: how often it ships, and which model types it bets on.
Compare model types
Filter by type: text/LLM, multimodal, image, video, or audio. The mix shows where the industry is pushing: text models still lead the count, while image and video generators ship in waves. Watch the balance shift month by month.
Catch up on what you missed
Page back through the timeline. Some months see dozens of launches, and the archive reaches back years. The board is a full release history, not just a feed of today's new AI models. Reconstruct any lab's year in a few scrolls.
New AI Models FAQ
What the board tracks, where the data comes from, and how to read it.
The board sorts releases newest first, so the top row is always the most recent launch from any provider. Recent months brought new AI models from labs like OpenAI, Anthropic, Google, Qwen and Zhipu AI. Every row links to the official announcement with its exact release date.
Twenty-nine providers and counting: OpenAI, Anthropic, Google, Meta, Microsoft, Amazon, NVIDIA, Mistral AI, DeepSeek, Qwen, Moonshot AI, Zhipu AI, ByteDance, xAI and more. Western and Chinese labs share one board, so a launch abroad still shows up the same day.
Five kinds: text/LLM, multimodal, image generation, video generation, and audio/speech. Text models carry the largest share of launches, but image and video generators ship in fast waves. The Type filter isolates any one stream.
Constantly. Major labs now ship new AI models on a weekly cadence, and busy weeks bring more than one frontier launch. Since 2024 the pace has kept climbing, and the board adds each release as its official announcement goes live.
Each row uses the date of the initial public release: the day the lab announced the model or opened it to users. Betas, rumors and paper preprints do not count. Every entry links to the official announcement, so you can verify the date yourself.
Yes. The board doubles as an archive reaching back to 2017. Page back to see how the release cadence exploded, from a handful of new AI models a year to hundreds. It is a searchable release history, not just a live feed.
Yes. The Provider filter narrows the board to one lab or any set of labs. Combine it with the Type filter to answer narrow questions, like every video model a single lab has shipped.
Yes, completely. Every release, filter and official link is free to browse, no signup needed. The board updates as new models are announced, so it pays to check back after every launch event.
Curated from official provider announcements; dates reflect the initial public release.