GPT-Live 1.1 from @OpenAI is on AI Gateway.
Compared with GPT-Live 1, it transcribes speech more accurately and keeps its accent and voice more consistent across languages.
ππππππ/πππ-ππππ-π·.π·
https://t.co/VL9095uWVP
eve 1.0 feels like my life work even though it was just a markdown file back in March. Iβm bias but I think itβs the best framework for building agents that lead to real business outcomes. And I donβt think itβs particularly close.
This is less to do with some brilliant light bulb moment and everything to do with being at a company that has access to the best builders in the world. We built this framework along side them. Every feature and API in eve went through many variations purely based on real world experience. Both internally working with different teams at Vercel and externally working with companies like Clay, Mux, Browserbase and many others.
Obviously any design can be copied and recreated in todayβs token economy. In many ways itβs very flattering to see it happen. But I donβt think the stream of feedback and constant obsession over meeting that feedback can be copied in the same way. Iβm very excited to keep delivering and see what people build.
https://t.co/hJtalv62gX
This one was super fun to build. You can now bring your own model to @vercel AI Gateway to centralize observability and access control for all your tokens.
My favorite part is that you can fall back to hundreds of catalog models in case your endpoint ever goes down--it's the best of both worlds.
Dedicated deployments, prerelease models, on prem, fine-tunes, the list goes on! Available on all plans :)
Private Inference is now on AI Gateway.
πππππππ/<ππππ>
Use your own fine-tuned, self-hosted, or prerelease models alongside catalog models, with the same APIs, auth, and logs.
shipped a big one today.
you can now jump into any subagent mid-task
- watch it work
- message it
- answer its prompts
it feels less like delegating and more like running a small team.
Parallel Decisions from @p0 is on AI Gateway, free until Nov 23.
A decision model that searches the web, returns a probability, and lists the citations alongside it.
ππππππππ/ππππππππ-πππππππππ-ππππ
https://t.co/Bdp1Y1Fsax
Vercel Token Pass now offers $150 in tokens for $30 per month.
β’ Use across the 5 models supported today
β’ Available in fx with no extra setup
https://t.co/LhHUi0Nuiz
GLM 5.3 Flash is now in Vercel Token Pass.
β’ Z.β ai's native multimodal coding model
β’ 1M-token context window
Get $80 of tokens for $30/mo β https://t.co/32yxnwpmT2
Search and Fetch tools from @deepseek_ai are live on AI Gateway.
Search the web and read pages with any tool-calling model. Switch model providers and your tools stay the same.
https://t.co/LEXQJovaoe
Great to see. A number of these games were formative, life-changing experiences for me, and are the clear ancestors of Baldurβs Gate III, Dragon Age, and other such RPGs. They also have some of my all-time favorite cover artwork.
Token Pass credit now applies automatically across AI Gateway.
Your usual API key, OIDC token, or fx login can use your Token Pass subscription credits for included models. Once they run out, API key and OIDC requests use team credits.
https://t.co/pFe6vQ4qCA
npm i -g ai-cli
Such a simple idea. An ππ CLI. Gives every agent access to video, voice, image, text generation. And ππ ππππππ brings Jev-class models.
All behind AI Gateway with cost and ZDR controls.
AI CLI
Generate images, video and audio from any agent, with any model
Claude Code + Opus + Flux
Codex + GPT + Gemini
opencode + DeepSeek + MAI
fx + GLM + Grok
npm i -g ai-cli
Behind the scenes of the Geist Serif animation βοΈ
Most of it was made in a custom motion tool Iβve been building on AI Gateway. The tool is still evolving, but itβs already changed how I work a lot. Crazy to see how fast motion design workflows are changing.