We are rolling out Advanced Voice to Plus and Team users over the next week. Voice conversations with ChatGPT are now much more natural. It has lower latency and you can interrupt if the responses are long. We also added support for memory to personalize the responses, new

realtime api (speech-to-speech): openai.com/index/introduc… vision in the fine-tuning api: openai.com/index/introduc… prompt caching (50% discounts and faster processing for recently-seen input tokens): openai.com/index/api-prom… model distillation (!!): openai.com/index/api-mode…

Launched at Dev Day today -- Real time API for low latency speech applications -- Vision fine-tuning -- Model distillation -- Prompt caching Look forward to seeing what developers build.

Try Canvas - a new way to collaborate with AI in ChatGPT to edit/refine your creations.

Chat Completions now supports audio input and output:

ChatGPT Search is live to plus and team users. You can get better answers with access to real-time information now.

12 days of OpenAI - we are going to launch a bunch of things - big and small. Some are still cooking - maybe a bumpy ride but it will be fun.

Day 1 launches: - o1: smarter, faster and multi-modal (image input). In chatgpt, coming to api soon. - ChatGPT Pro: unlimited usage and o1 smarter mode that thinks longer

Day 2 of 12 days of OpenAI: Reinforcement Fine-tuning: Create custom models that improve reasoning performance for your own tasks with very little data. Alpha launch now. Will be available widely in q1.

Day 3 of 12 days of OpenAI sora.com - AI powered video generation will enable entirely new forms of creative expression. Amazing work by the Sora team! Can’t wait to see what people create.

Day 4 of 12 days of OpenAI Canvas is now available to everyone. You can run Python code inside Canvas and bring Canvas to Custom GPTs. Canvas is a more collaborative human+AI interaction to write and code in ChatGPT. Look forward to seeing what you create.

Day 5 of 12 days of OpenAI ChatGPT integration into Apple Intelligence is now GA. Key features: - Camera Control: Ask ChatGPT anything about what you are looking at with a single tap - Siri: Ask any question/search the web to ChatGPT via Siri - Writing Tools: Compose

Day 7 of 12 days of OpenAI Projects in ChatGPT: You can organize your chats into specific folders and add custom instructions or files that are specific to that project for a better experience.

Day 8 of 12 days of OpenAI: ChatGPT Search Improvements — Launching to all logged in (including free) users — Search with advanced voice mode — Faster navigation and better mobile experience including maps integration

Day 9 of 12 days of OpenAI Today is all about developers - o1 in available in the API ---- Support for function calling, structured outputs and developer messages to tune model behavior. ---- Vision input ---- A new reasoning_effort parameter to control how much time the model

Day 10 of 12 days of OpenAI Access ChatGPT on WhatsApp or via a phone call (1-800-ChatGPT) We want to make ChatGPT available as broadly as possible in any communication medium of your choice. I love that this started as a fun hackathon project a few weeks ago.

Day 12 of 12 days of OpenAI Our next frontier model: o3 and o3-mini. Now open for safety testing (not a general launch yet). It is a big step forward in math, science, coding and reasoning abilities, beating human expert level performance in many of these benchmarks.

Day 12 of 12 days of OpenAI Our next frontier model: o3 and o3-mini. Now open for safety testing (not a general launch yet). It is a big step forward in math, science, coding and reasoning abilities, beating human expert level performance in many of these benchmarks.