Refresh
The next step is fine-tuning, and the API model currently supports preference and supervised types. So, for the ninth day, OpenAI ushered in the o1 model into the API, along with real-time improvements and tweaks.
A few minor items also shipped today – including new SDKs – Go and Java -, an updated flow for creating an API key, OpenAI dev talks posted on YouTube, and an AMA, as of now, for the next hour on the OpenAI developers forum.
And the ninth day is over. It was still fast, although a little longer than day eight, but we’ll have to see what day ten has in store for us tomorrow. OpenAI hasn’t announced what’s coming, which means everything is on the table.
Some good news is that o1 uses 60% fewer tokens than the o1 preview, which should mean you get even more value from the o1 version, which is rolling out today – although it will take a few weeks to that it affects all users. .
That’s not all, though, OpenAI is now aimed at real-time AI, which is all about creating voice experiences. Currently it supports web sockets, but webRTC support is here and is designed for the Internet.
Now, rather than an elf on the shelf, OpenAI has released a fawn on the lawn with a microcontroller inside. When the Fawn was plugged into the MacBook Pro it started saying “Merry Christmas, what are you talking about?” and responded by saying that webRTC was a bit complicated, and Fawn shifted the conversation to giving gifts at Christmas. The brief here is that you can create custom assistants using voice – pretty cool.
It’s time for day nine of the 12 days of OpenAI, and as announced during day eight, we’re expecting a sort of mini-developer day. And right away it’s confirmed that it’s all about how developers and startups can build on the OpenAI API.
To start, o1 is launching today in preview in the API and, based on the return function calls, messages from the developer and structured output arrive with the model. OpenAI will launch a “reasoning effort” tool within the o1 model, which will allow you to tell the AI how much time to spend on a task.
Finally, the o1 model has a vision aspect: in the example, it examines a tax form and can help detect problems in the form. To do this, OpenAI has opened its developer program and, via a “developer message”, you can provide instructions on how it works. This was written, the tax forms were uploaded and the AI analyzed the image and marked the errors. o1 here analyzes the uploaded items and provides written responses.
ChatGPT and Perplexity search
With OpenAI releasing some updates to the ChatGPT search feature yesterday, our Senior AI Editor Graham thought it would be a good time to compare what ChatGPT’s AI search feature can do. against perplexitywhich has been doing AI searches for a while now. He was quite surprised by the results. Read and see what you think.
Today’s more developer-focused announcement will take place at 10:00 a.m. PT.
Operator agent
With many AI agents (autonomous AI robots that execute your bids) waiting to become a key feature of 2025, we’re still wondering if we’ll see the release of Operator from OpenAI this week. The smart money is here to close the show on Friday.
The operator would be able to control your browser, meaning it could perform whatever task you’re currently using your browser for, which, when you think about it, is probably about 90% of what you’re currently using your browser for. use your computer.
The operator is already expected to land in January, so it’s not out of the realm of possibility that OpenAI has moved the release forward by a few weeks.
Welcome to day nine!
Welcome to day nine of 12 Days of OpenAI! Yesterday was a real Christmas cracker, since we had Search ChatGPT available to all ChatGPT users, including those on the free tier. The OpenAI team has hinted that today’s announcement will be more developer-focused, but check back at 10:00 AM PT and we’ll have the news for you as it happens.
As quickly as day 8 of 12 Days of OpenAI began, it is already coming to an end. ChatGPT Search was the center of attention, with significant improvements and a much larger rollout to free, logged-in users worldwide where ChatGPT is available. Just like Canvas, you’ll need a free account to use ChatGPT search and benefit from high rate limits.
Kevin also announced that tomorrow, Day 9, would be a mini developer day, so expect the focus to be less on mainstream features and more on bigger tools.
Beyond rolling out ChatGPT search to more users, OpenAI is also integrating the feature more seamlessly into its Android or iOS mobile app. When you ask a question, for example about a restaurant in a specific area, like the OpenAI team demonstrated during the reveal, the results will be listed online. Plus, you can have a more natural conversation about the results to find what you’re really looking for. It’s pretty cool.
Once you find a restaurant in the ChatGPT app for iOS, you can also get directions through Apple Maps as it is integrated.
Additionally, in the mobile app, you can speak with ChatGPT in voice mode, and it will incorporate search results and broader web information into its response. For example, if you ask for a Christmas market, you can even be more specific about opening hours and days.
It’s all about ChatGPT research for day 8
Here we go ! OpenAI kicks off day eight of its 12 days of announcements. OpenAI Product Manager Kevin kicks things off and quickly shared that the focus today is on ChatGPT Search.
First, it’s coming to everyone around the world and on all platforms where ChatGPT is available starting today. OpenAI also generally states that it has improved ChatGPT search and is rolling out the ability to search while you speak with ChatGPT’s advanced voice mode.
We asked ChatGPT what announcement we will receive today
We asked ChatGPT what he thought OpenAI would be released today because, well, if anyone was going to know, it would have to be ChatGPT, right? He came back with:
“Given that image generation updates have been noticeably absent thus far, many believe that a DALL-E update could come today. The announcement is scheduled for 10 a.m. PT, so keep an eye out for news regarding potential advancements to OpenAI’s authoring tools and accessibility features.
To be honest, we think that’s true, but it also sounds a lot like ChatGPT read our own blog post on the topic of today’s release (see below), so, uh, thanks for nothing ChatGPT…
Time to create an AI podcast generator?
Something we haven’t seen from OpenAI so far is an AI podcast generator. Although Google has had great success with NotebookLM, its search tool that will generate an incredibly real podcast between two AI hosts from the text, video, or PDF sources you feed it, we haven’t seen anything from OpenAI on this front, therefore far.
Google deploys a new NotebookLM feature it also allows you to participate in the conversation with the AI hosts.
NotebookLM has been so popular that we’re starting to wonder if this week we’ll see OpenAI enter the AI podcast game with one of its “12 Days of OpenAI” releases? Time will tell.
Let’s talk about AI images
Could we see a major DALL-E update in today’s announcement? I highly doubt it, but you never know.
Today, TechRadar Senior AI Editor John-Anthony Disotto tested Grok, a DALL-E competitor from xAI, Elon Musk’s AI company. Grok 2 is now free for all users on X (formerly Twitter), and it is capable of generating crazy image generation results without restrictions.
In his article entitled “I used the new free Grok tier on X but I can’t show you the results because it might infringe Nintendo’s copyrights“, he explains that OpenAI’s AI image generator will not create images of copyrighted people or public figures, while Grok will do whatever you ask it to do. Despite the limitations, DALL -E 3 as part of ChatGPT remains one of our choices for best AI image generatorbut could it get much better with today’s announcement? Time will tell.
My favorite announcement from last week?
Today is a good day to reflect on the freebies OpenAI announced last week, and while Sora was the highlight, it was definitely another of its announcements that I found the most useful…
I’m talking about the Canvas feature. As I wrote at the time, he completely changed the way I use ChatGPT. The new writing tools are really useful, and I like how you can keep refining the same piece of text over and over again, without having to generate reams of text every time you want to change a single piece of it .
If you haven’t played with Canvas yet, I recommend you give it a try. It’s free!
Welcome to day eight of “12 Days of OpenAI”! We’ve had the weekend to reflect on all the good stuff OpenAI released last week (Sora, ChatGPT Canvas and Projects, plus ChatGPT o1) and it makes us wonder what we can expect from the IT giant. AI this week?
There’s still nothing announced regarding AI image generation, so could we see a new version of DALL-E today?
We’re here from today until Friday of this week (the 20th), when we’ll have our last day of OpenAI releases. Today’s announcement begins at 10:00 AM PT, so don’t miss it.
