Please note: Most, if not all, of the articles published at this website were completed by Chat GPT (chat.openai.com) and/or copied and possibly remixed from other websites or Feedzy or WPeMatico or RSS Aggregrator or WP RSS Aggregrator. No copyright infringement is intended. If there are any copyright issues, please contact: bicycledays@yahoo.com.
With the vacation season upon us, many corporations are discovering methods to take benefit via offers, promotions, or different campaigns. OpenAI has discovered a solution to take part with its “12 days of OpenAI” occasion sequence.
OpenAI introduced by way of an X put up that beginning on Dec. 5, the corporate would host 12 days of dwell streams and launch “a bunch of recent issues, huge and small,” in line with the put up. The corporate saved one of the best for final, sharing its greatest annoouncemnt on Friday, Dec. 20, the final day of the sequence.
Here is the whole lot it’s good to know in regards to the marketing campaign, in addition to a round-up of day by day’s drops.
What are the ’12 days of OpenAI’?
OpenAI CEO Sam Altman shared extra particulars in regards to the occasion, which kicked off at 10 a.m. PT on Dec. 5 and occurred day by day for 12 weekdays with a dwell stream that includes a launch or demo. The launches included each “huge ones” or “stocking stuffers,” in line with Altman.
🎄🎅beginning tomorrow at 10 am pacific, we’re doing 12 days of openai. every weekday, we could have a livestream with a launch or demo, some huge ones and a few stocking stuffers. we’ve bought some nice stuff to share, hope you get pleasure from! merry christmas.
On the final day of OpenAI, OpenAI unveiled its newest fashions, o3, which embody o3 and o3 mini.
As beforehand reported, the explanation for the identify that skips “o2” is the existence of Telefonica’s O2 telecommunications model, which might trigger confusion and copyright points.
OpenAI stated that the know-how won’t be obtainable to most of the people but.
o3 can outperform o1 in quite a lot of benchmarks, together with math and science, as seen in its efficiency on the AIME 2024, a contest math benchmark, and the GPQA, a Ph.D. degree science benchmark with biology, physics, and chemistry questions.
o3 additionally scored a brand new state-of-the-art rating on the ARC-AGI benchmark, which is critical as a result of it reveals the mannequin is nearing AGI, though, to be clear, it’s not there but.
o3 mini is a brand new mannequin within the o3 household that may help three reasoning choices: low, medium, and excessive. The pondering time determines the efficiency. At a low degree, it performs the identical as o1-mini on the Codeforces Competitors Code benchmark, however at medium and excessive, it might probably carry out corresponding to o1, if not higher. This efficiency remained constant throughout different evals.
Along with inner security testing, OpenAI is, for the primary time, opening up the o3 fashions to exterior security testing. Security researchers can get early entry to the mannequin by filling out a kind on the OpenAI web site, which can be open till January 10.
Sam Altman concluded the dwell stream by sharing that the o3 mannequin is deliberate to launch on the finish of January, and the total o3 mannequin can be launched after that.
The corporate additionally launched deliberative alignment, “a coaching paradigm that instantly teaches reasoning LLMs the textual content of human-written and interpretable security specs, and trains them to purpose explicitly about these specs earlier than answering,” in line with OpenAI.
Thursday, December 19
On the second to final day of ’12 days of OpenAI,’ the corporate centered on releases concerning its MacOS desktop app and its interoperability with different apps.
Customers can now use the desktop app on MacOS to see and automate their work with ChatGPT. There can be extra releases of this nature in 2025, however till then, OpenAI has been introducing the three options under.
Utilizing the “Work with Apps” button, customers can now work with many extra coding apps. The record contains: BBEdit, MatLab, Nova, Script Editor, TextMate, Android Studio, AppCode, CLion, DataGrip, GoLand, IntelliJ IDEA, PHPStorm, PyCharm, RubyMine, RustRover, WebStorm, Immediate, and Warp.
For customers who use ChatGPT for writing, the desktop app now helps Apple Notes, Quip, and Notion.
Lastly, the desktop app for MacOS now helps Superior Voice Mode whereas working with different apps.
Options have already been shipped. All it’s a must to do is have the most recent model of the MacOS app and a Plus, Professional, Group, Enterprise, and Edu subscription.
To ease privateness considerations, OpenAI says ChatGPT will solely work with apps when manually prompted. When the function is lively, customers know what can be hooked up to the message.
“Day 12, now we have one thing tremendous particular, so do not miss it,” teased OpenAI about its upcoming Friday launch.
Wednesday, December 18
Have you ever ever wished to make use of ChatGPT with no Wi-Fi connection? Now, all it’s a must to do is place a cellphone name. Here is what OpenAI launched on the tenth day:
By dialing 1-800-ChatGPT, now you can entry the chatbot by way of a toll-free quantity. OpenAI encourages customers to save lots of ChatGPT of their contacts for straightforward entry.
Customers can name anyplace within the US; in different international locations, customers can message ChatGPT on WhatsApp. Customers get quarter-hour of free ChatGPT calls per thirty days.
In WhatsApp, customers can enter a immediate by way of a textual content as they might with every other particular person of their contacts. On this expertise, it’s only a textual content message.
The cellphone name function works on any cellphone, from a smartphone to a flip cellphone — even a rotary cellphone.
The presenters stated it’s meant to make ChatGPT extra accessible to extra customers.
Tuesday, December 17
The releases on the ninth day all give attention to developer options and updates, dubbed “Mini Dev Day.” These launches embody:
The o1 mannequin is lastly out of preview within the API with help for operate calling, structured outputs, developer messages, imaginative and prescient capabilities, and decrease latency, in line with the corporate.
o1 within the API additionally includes a new parameter: “reasoning effort.” This parameter permits builders to inform the mannequin how a lot effort is put into formulating a solution, which helps with value effectivity.
OpenAI additionally launched WebRTC help for the Realtime API, which makes it simpler for builders “to construct and scale real-time voice merchandise throughout platforms.”
The Realtime API additionally bought a 60% audio token value drop, help for GPT-4o mini, and extra management over responses.
The fine-tuning API now helps Choice Wonderful-Tuning, which permits customers to “Optimize the mannequin to favor desired conduct by reinforcing most popular responses and decreasing the chance of unpreferred ones,” in line with OpenAI.
OpenAI additionally launched new Go and Java SDKs in beta.
An “AMA” (ask me something) session can be held for an hour after the dwell stream on the OpenAI GitHub platform with the presenters.
Monday, December 16
The drops for the second Monday within the 12 days of OpenAI sequence all centered on Search in ChatGPT.
The AI search engine is accessible to all customers beginning as we speak, together with all free customers who’re signed in anyplace they will entry ChatGPT. The function was beforehand solely obtainable to ChatGPT Plus customers.
The search expertise, which permits customers to browse the online from ChatGPT, bought quicker and higher on cell and now has an enriched map expertise. The upgrades embody image-rich visible outcomes.
Search is built-in into Advance Voice mode, which means now you can search as you discuss to ChatGPT. To activate this function, simply activate Advance Voice the identical manner you recurrently would and ask it your question verbally. It’s going to then reply your question verbally by pulling from the online.
OpenAI additionally teased builders, saying, “Tomorrow is for you,” and calling the upcoming livestream a “mini Dev Day.”
Friday, December 13
Certainly one of OpenAI’s most extremely requested options has been an organizational function to raised preserve monitor of your conversations. On Friday, OpenAI delivered a brand new function referred to as “Initiatives.”
Initiatives is a brand new solution to manage and customise your chats in ChatGPT, meant to be part of persevering with to optimize the core expertise of ChatGPT.
When making a Venture, you possibly can embody a title, a custom-made folder shade, related mission information, directions for ChatGPT on the way it can finest aid you with the mission, and extra in a single place.
Within the Venture, you can begin a chat and add earlier chats from the sidebar to your Venture. It may possibly additionally reply questions utilizing your context in a daily chat format. The chats will be saved within the Venture, making it simpler to select up your conversations later and know precisely what to search for the place.
It is going to be rolled out to Plus, Professional, and Groups customers beginning as we speak. OpenAI says it is coming to free customers as quickly as doable. Enterprise and Edu customers will see it rolled out early subsequent yr.
Thursday, December 12
When the dwell stream began, OpenAI addressed the elephant within the room — the truth that the corporate’s dwell stream went down the day earlier than. OpenAI apologized for the inconvenience and stated its workforce is engaged on a autopsy to be posted later.
Then it bought straight into the information — one other highly-anticipated announcement:
Superior Voice Mode now has screen-sharing and visible capabilities, which means it might probably help with the context of what it’s viewing, whether or not that be out of your cellphone digicam or what’s in your display screen.
These capabilities construct on what Superior Voice might already do very effectively — participating in informal dialog as a human would. The natural-like conversations will be interrupted, have multi-turns, and perceive non-linear trains of thought.
Within the demo, the person will get instructions from ChatGPT’s Superior Voice on make a cup of espresso. Because the demoer goes via the steps, ChatGPT is verbally providing insights and instructions.
There’s one other bonus for the Christmas season: Customers can entry a brand new Santa voice. To activate it, all customers need to do is click on on the snowflake icon. Santa is rolling out all through as we speak in every single place that customers can entry ChatGPT voice mode. The primary time you discuss to Santa, your utilization limits reset, even if in case you have reached the restrict already, so you possibly can have a dialog with him.
Video and display screen sharing are rolling out within the newest cell apps beginning as we speak and all through subsequent week to all Group customers and most Professional and Plus subscribers. Professional and Plus subscribers in Europe will get entry “as quickly as we will,” and Enterprise and Edu customers will get entry early subsequent yr.
Wednesday, December 11
Apple launched iOS 18.2 on Wednesday. The discharge contains integrations with ChatGPT throughout Siri, Writing Instruments, and Visible Intelligence. Because of this, the dwell stream centered on strolling via the combination.
Siri can now acknowledge once you ask questions outdoors its scope that would profit from being answered by ChatGPT as a substitute. In these situations, it is going to ask if you would like to course of the question utilizing ChatGPT. Earlier than any request is shipped to ChatGPT, a message notifying the person and asking for permission will all the time seem, putting management within the person’s fingers as a lot as doable.
Visible Intelligence refers to a brand new function for the iPhone 16 lineup that customers can entry by tapping the Digital camera Management button. As soon as the digicam is open, customers can level it at one thing and search the online with Google, or use ChatGPT to be taught extra about what they’re viewing or carry out different duties corresponding to translating or summarizing textual content.
Writing Instruments now includes a new “Compose” device, which permits customers to create textual content from scratch by leveraging ChatGPT. With the function, customers may even generate photographs utilizing DALL-E.
The entire above options are topic to ChatGPT’s day by day utilization limits, the identical manner that customers would attain limits whereas utilizing the free model of the mannequin on ChatGPT. Customers can select whether or not or to not allow the ChatGPT integration in Settings.
Learn extra about it right here: iOS 18.2 rolls out to iPhones: Attempt these 6 new AI options as we speak
Tuesday, December 10
Canvas is coming to all net customers, no matter plan, in GPT-4o, which means it’s not simply obtainable in beta for ChatGPT Plus customers.
Canvas has been constructed into GPT-4o natively, which means you possibly can simply name on Canvas as a substitute of getting to go to the toggle on the mannequin selector.
The Canvas interface is identical as what customers noticed in beta in ChatGPT Plus, with a desk on the left hand aspect that reveals the Q+A alternate and a right-hand tab that reveals your mission, displaying the entire edits as they go, in addition to shortcuts.
Canvas will also be used with customized GPTs. It’s turned on by default when creating a brand new one, and there may be an possibility so as to add Canvas to present GPTs.
Canvas additionally has the power to run Python code instantly in Canvas, permitting ChatGPT to execute coding duties corresponding to fixing bugs.
Learn extra about it right here: I am a ChatGPT energy person – and Canvas remains to be my favourite productiveness function a month later
Monday, December 9
OpenAI teased the third-day announcement as “one thing you’ve got been ready for,” adopted by the much-anticipated drop of its video mannequin — Sora. Here is what it’s good to know:
Generally known as Sora Turbo, the video mannequin is smarter than the February mannequin that was previewed.
Entry is coming within the US later as we speak; customers want solely ChatGPT Plus and Professional.
Sora can generate video-to-video, text-to-video, and extra.
ChatGPT Plus customers can generate as much as 50 movies per thirty days at 480p decision or fewer movies at 720p. The Professional Plan affords 10x extra utilization.
The brand new mannequin is smarter and cheaper than the previewed February mannequin.
Sora options an discover web page the place customers can view one another’s creations. Customers can click on on any video to see the way it was created.
A dwell demo confirmed the mannequin in use. The demo-ers entered a immediate and picked facet ratio, period, and even presets. I discovered the dwell demo video outcomes to be life like and beautiful.
OpenAI additionally unveiled Storyboard, a device that lets customers generate inputs for each body in a sequence.
Friday, December 6:
On the second day of “shipmas,” OpenAI expanded entry to its Reinforcement Wonderful-Tuning Analysis Program:
The Reinforcement Wonderful-Tuning program permits builders and machine studying engineers to fine-tune OpenAI fashions to “excel at particular units of complicated, domain-specific duties,” in line with OpenAI.
Reinforcement Wonderful-Tuning refers to a customization method by which builders can outline a mannequin’s conduct by inputting duties and grading the output. The mannequin then makes use of this suggestions as a information to enhance, turning into higher at reasoning via related issues, and enhancing total accuracy.
OpenAI encourages analysis institutes, universities, and enterprises to use to this system, significantly those who carry out slim units of complicated duties, may benefit from the help of AI, and carry out duties which have an objectively appropriate reply.
Spots are restricted; candidates can apply by filling out this kind.
OpenAI goals to make Reinforcement Wonderful-Tuning publicly obtainable in early 2025.
Thursday, December 5:
OpenAI began with a bang, unveiling two main upgrades to its chatbot: a brand new tier of ChatGPT subscription, ChatGPT Professional, and the total model of the corporate’s o1 mannequin.
The complete model of o1:
Shall be higher for all types of prompts, past math and science
Will make main errors about 34% much less typically than o1-preview, whereas fascinated by 50% quicker
Rolls out as we speak, changing o1-preview to all ChatGPT Plus and now Professional customers
Lets customers enter photographs, as seen within the demo, to supply multi-modal reasoning (reasoning on each textual content and pictures)
ChatGPT Professional:
Is supposed for ChatGPT Plus superusers, granting them limitless entry to one of the best OpenAI has to supply, together with limitless entry to OpenAI o1-mini, GPT-4o, and Superior Mode
Options o1 professional mode, which makes use of extra computing to purpose via the toughest science and math issues
Prices $200 per thirty days
The place are you able to entry the dwell stream?
The dwell streams had been held on the OpenAI web site, and posted to its YouTube channel instantly after. So for those who missed the 12 days of OpenAI and need to rewatch, you possibly can entry all of them on the corporate’s YouTube channel.
