@orotoi1 no I meant if there's an update its nice to have some video showcasing the features. Just on the day of the release even if that release is say months later
Thank God! This is one of the 12 gifts I was begging for. I am a fiction writer, and I use Canvas 95% of the time when I use ChatGPT. This is a godsend. Now, if you can add advanced voice with Canvas, OMG, it would be a dream come true. 🥰
@@Necca_UK Of course not. 'Writers' who are having to use ChatGPT, other than just an editing tool and brainstorming, don't know what good writing is, and will flood Amazon with crud. ChatGPT overwrites/writes like an beginning writer. It will no doubt get better, but at the moment it's poor.
What a time to be alive. Absolute geniuses, under the cloak of simplicity, explain what they have been up to in the last months. This is just so amazing.
Awesome! One thing, could you add folders or an organization system for chats and projects? Other sites like Perplexity or NotebookLM let you create 'spaces' that let you organize all the chats for one project
That 12 days of OpenAI is so amazing on so many levels, one of them is to really get to you guys at OpenAI, I love this! And BOY, will I do incorporate Canvas into my current GPT I am working with, this gonna be awesome!
Dear OpenAI Team, I suggest considering some Whisper integration in the browser. Using voice instead of typing is much more comfortable. I use Wispr Flow all day.
Месяц назад+10
Thank you! I'd not heard of this and spent HOURS trying to get ChatGpt to understand what I was trying to dictate. Will purchase. If only it did Math.
I've always found the idea of talking into your device incredibly awkward, not to mention it's not always accurate to begin with. Maybe it's my nerd side showing but I'd rather just use a keyboard.
Canvas is a fantastic feature that will greatly enhance my work. I look forward to the addition of project creation and project knowledge capabilities, enabling ChatGPT to comprehend and continue work within specific projects, with dedicated chats for each. Additionally, it would be beneficial if ChatGPT could remember content from other project-related chats to avoid the need for copying and pasting when reaching message or context length limits. This would streamline workflows and improve efficiency. Thank you, OpenAI!
@@thejeffyb9766 On the contrary, my kids love interacting with advanced voice mode to generate stories, prompting and tweaking as we go. Previously id get frustrated reading to them and having them interrupt, but the interactivity makes it enjoyable for all of us.
Highlights: - Launch of Canvas for all users, integrating it into the main ChatGPT model. - Ability to run Python code directly within Canvas, providing instant feedback. - New side-by-side view for easier collaboration and editing. - Features include commenting on essays and debugging code with ChatGPT’s assistance. - Custom GPTs can now leverage the power of Canvas for enhanced functionality.
They‘ll surely update the advanced voice mode in the next couple of days, but I dont know, if they integrate file reading with it. I rather thought of the vision feature.
@@omegangerman I just want advanced voice mode to work with files cuz as a student. The advance voice mode really helps a lot cuz the conversations are just seamless. With the normal voice mode it takes like 6 second to wait for a response
Just a thought. I think OpenAI team members are really used to work with much better in-house version of these tools, they seemed genuinely confused and nervous when the public ChatGPT version just didn't do the task right at the first try during live demonstration 😁 The canvas tool has been probably already undergoing red teaming with O1 and who knows what else on board internally for the last few months at least.
As a developer, that's often the experience. You spend days preparing for a demo and everything works perfectly. Then, when it finally comes time for the real thing, it crashes, runs slowly, or doesn't run at all. It's frustrating and embarrassing and happens to just about everybody.
I've see now that Canvas is now available in the Windows desktop app. I've also noted that the OSx version can integrate into VS Code... I'm hoping that us Windows users get this integration feature soon.
I have been using canvas for a while, and I think it is great. These improvements will help. However, it does become limited because it has a relatively short context. Once you get over a couple of hundred lines of code or maybe 1500 words, it becomes very unreliable. It will loose part of your edit or refuse to make changes and gas light you to say that it has made changes. Can you please, please please increase the context length significantly of the canvas.
I have the same problem. I write blog posts that are over 2,000 words long and stories that are much longer. For shorter content, I've been very happy with canvas. For longer works, the feature is largely unusable at the moment.
*Canvas Updates - Enhanced Collaboration and Functionality within ChatGPT* * *0:12** Introduction:* Kevin Weil introduces Day 4 of "12 Days of OpenAI," focusing on Canvas updates. * *0:31** Canvas Launch:* Canvas, previously in beta for Plus users, is now available for all users and integrated into the main ChatGPT model. It facilitates collaborative writing and coding beyond simple chat. * *0:49** Key Features:* Three key updates are highlighted: 1) General availability of Canvas, 2) Python code execution within Canvas with visual output, and 3) Canvas integration with custom GPTs. * *1:20** Collaborative Storytelling:* Lee demonstrates creating a Christmas story using Canvas, showcasing the side-by-side view for chat and document editing, collaborative editing features, and shortcut buttons for suggesting edits, adjusting length, refining reading level, polishing, and adding emojis. * *5:25** Essay Feedback:* Alexi demonstrates using Canvas for essay feedback, highlighting the in-line commenting feature, which allows ChatGPT to provide targeted suggestions. The apply/dismiss buttons allow users to accept or reject proposed edits. The copy button facilitates easy transfer of the finalized text. * *8:30** Programming Enhancements:* Canvas now supports Python code editing with syntax highlighting and autocomplete. A "Run" button enables code execution within Canvas, providing instant feedback. * *9:13** Debugging with Canvas:* Alexi demonstrates debugging Python code in Canvas, showing how ChatGPT identifies errors, provides corrected code, and allows users to view changes. * *11:17** Data Visualization:* Alexi demonstrates generating graphics (line chart and Sankey diagram) within Canvas by running Python code, showing how ChatGPT can assist with data visualization tasks. * *13:38** Custom GPTs Integration:* Canvas is now integrated with custom GPTs, enabling users to incorporate Canvas functionality into their specialized GPTs. * *14:00** Santa GPT Example:* Alexi demonstrates a custom GPT designed for Santa to draft letter responses, showcasing how Canvas can be automatically triggered based on custom instructions. * *16:50** Custom GPT Configuration:* The process of configuring a custom GPT is shown, highlighting the instructions section where Canvas usage can be specified and the capabilities checkboxes for enabling Canvas functionality. * *18:14** Rollout and Availability:* Canvas is rolling out to all web users, free and paid. Existing GPTs will need to manually enable Canvas through the configuration settings. I used gemini-exp-1121 on rocketrecap dot com to summarize the transcript. Cost (if I didn't use the free tier): $-0.0211 Input tokens: 20609 Output tokens: 522
Honestly, one of the better ideas for a feature.... is that when you start a new chat, chatGPT could use the context of this new conversation to perform some semantic similarity to previous conversations. The idea being that you may have had a conversation in the past that you have forgotten about..... but it relates heavily to what you are currently engaging chatgpt about. Or, that you know you spoke with chatGPT about it before, but instead of searching through the hundreds if not thousands of individual threads, chatgpt could just relate your new inquiry in some seamless fashion.... for example: "Oh yeah, I remember you talking about that back in September.... we spoke about x,y,z at length. Would you like a very brief summary of that thread?" A smart search if you will that is chat aligned and natural feeling.
In this video you guys show that the feature is super buggy, it hangs, code does not work, when it does not throw errors the resulting pic does not make sense at all… the face of Kevin tell everything I need to know of the readiness of Canvas
I'm foreign English teacher in Taiwan being one of those like me, carries a lot of challenges, but hey, thanks for this awesome updates, teacher like me has a teaching assistant that gets the job done. Congratulations!
@boevayasemga problem is, the right side readjusts to add even more white space. Shrinks the content area instead of filling it fully. There's like 2 sizes, a smaller and a larger width. It snapping to the smaller width is what I have a problem with. Too much white space when it should just fully fill the area at those sizes.
I wish OpenAI made a Cursor competitor. There really is no substitution for having context within the IDE, and copy-pasting gets annoying for larger projects.
Ketvirta diena tikrai įspūdinga! Bet būtų smagu, jei tekstų kiekis, su kuriais reikia bendrauti, būtų šiek tiek mažesnis - kartais per daug apkrauna. Vis dėlto, labai įdomu stebėti, kaip OpenAI juda į priekį!
Canvas needs more language support but this is great start with Python. Though would have preferred to start with HTML/CSS/JS & React like Claude Artifacts.
Yeah, but how many projects we work on are a single file? I've got projects with multiple modules. This is where Anthropic gets it with their "Projects". GhatGPT just feels like a neat toy for developing smaller custom tooling, as opposed to a tool to accomplish real-world work.
• o1 model - data analytics / interpretation (technical & scientfic field) • Sora - visual rendering (creative industry like special visual effects) • Canvas - text writing (can ChatGPT differentiate whether what it wrote is factual based or fictional creation ?) When Canvas came out with pages of draft (be it for movie screenplay, film scriptwriting or best selling novel), would it be able to grade the content it produced will be a blockbuster success or a bestseller event when it gets 'published' out to the market (maybe to an 80% accuracy rate) ? → just an attempt to gauge its commercial viability upon its 'democratized' adoption out there.
How about the following things (that I've seen popping up here): - Enable advanced voice mode, even when interacting with uploaded files, graphics, or input text. Then we could prompt how to interact with the model. Would this be possible? - Integrate Whisper for voice interactions in the browser, similar to the desktop app's capabilities. - Add folders or enhanced options for organizing chats, projects, and prompts. - Allow the use of o1 or o1-mini in Custom GPTs for improved functionality.
Be it online search or Canvas, with the new features that are being added, I still lack the function to deactivate them if necessary without having to write it down every time
The Canvas preview has been a game changer for me - and being completely honest, is the main reason I kept my subscription with OpenAI as I've been very tempted by switching to the competition. Python in canvas is interesting, but the real power will be integrating Canvas with users IDE of choice. At first I'm thinking "If only I could run .mjs in Canvas..." but in reality, if only Canvas was integrated directly with my VS Code node.js session. I'm lead to believe that the OSx version of ChatGPT may do this, so I guess my question is when will we see this direct canvas integration feature for Windows/VS Code?
Didn’t we basically have this feature already lmao… I was thinking that shipmas for 12 days eventually it wouldn’t be very interesting… didn’t think it would be in the 4th day.
Can you please increase the context window so we can upload abt 5000 lines of code and edit in the Canvas? That way we don’t have to write code in stages. Regular context size makes the system crash before completion.
Είμαι τόσο χαρούμενος, γιατί μου θυμίζετε τρεις παλιούς μου φίλους εδώ στη Mύκονο στην Eλλάδα. Tον Bασίλη, τον Tσόπελα και τον Απόλλωνα... Τον καιρό του "Bolero Bar"
I can't get into SORA either, if your preparation wasn't good enough for issues during a SORA launch, what what will happen when you push out AGI. "We’re currently experiencing heavy traffic and have temporarily disabled AGI safety protocols. If you've never filled out your will before, please fill it out soon."
Thank you for the really new feature : code execution. But could you give us a Canvas that produces and edits an HTML + JavaScript + CSS page with code execution ? It’s the kind of single-page application that would be useful for everyone. (Python is just one option among other languages.) Editing an HTML page containing JavaScript shouldn’t be that difficult and seems equally universal in terms of usage.
Will Csnvas run with Blender (open source with unique flavor of Python) or other 3D spp, or can I link to run the code via the built-in or another text editor?
OpenAI please continue to make these videos into the future even after the shipmas days. They are very helpful
that means they will have to release something new every day xD
@orotoi1 no I meant if there's an update its nice to have some video showcasing the features. Just on the day of the release even if that release is say months later
+1 for feature walkthroughs, but also, 365 days of AI is what I'm here for.
Hopefully not with a $200 license everyday.
@@ricardocarreon I would pay $200 a day if it meant singularity in a week. 😆
I just tried Canvas for the first time. Mind Blown. Congratulations OpenAI (you must be so proud). This tool will transform the way I work. Thank you!
Thank God! This is one of the 12 gifts I was begging for. I am a fiction writer, and I use Canvas 95% of the time when I use ChatGPT. This is a godsend. Now, if you can add advanced voice with Canvas, OMG, it would be a dream come true. 🥰
Do you feel under threat?
@@Necca_UK what the hell dym?
@@Necca_UK Of course not. 'Writers' who are having to use ChatGPT, other than just an editing tool and brainstorming, don't know what good writing is, and will flood Amazon with crud. ChatGPT overwrites/writes like an beginning writer. It will no doubt get better, but at the moment it's poor.
this almost like movie HER
Use an app called Cursor
What a time to be alive. Absolute geniuses, under the cloak of simplicity, explain what they have been up to in the last months. This is just so amazing.
Now this is actually a great feature i will use for work. For me this is much bigger then Sora.
Lol
the software improvements are very good
Lmao why are you comparing this to sora?
Same!
I agree. I am a fiction writer, and it would be nice for me to be able to upload a story and have it read parts for characters. 🥰
OpenAI has a very specific focus on not just AI, but also UX. That's driving a lot of their value, and gets lets talked about.
Awesome!
One thing, could you add folders or an organization system for chats and projects? Other sites like Perplexity or NotebookLM let you create 'spaces' that let you organize all the chats for one project
Love NotebookLM
Absolutely
We really need this!
We really need this!
Great idea! This would also be a great way to organize memory context as well. Let's see it!
That 12 days of OpenAI is so amazing on so many levels, one of them is to really get to you guys at OpenAI, I love this! And BOY, will I do incorporate Canvas into my current GPT I am working with, this gonna be awesome!
Dear OpenAI Team, I suggest considering some Whisper integration in the browser. Using voice instead of typing is much more comfortable. I use Wispr Flow all day.
Thank you! I'd not heard of this and spent HOURS trying to get ChatGpt to understand what I was trying to dictate. Will purchase. If only it did Math.
In dutch please
I've always found the idea of talking into your device incredibly awkward, not to mention it's not always accurate to begin with. Maybe it's my nerd side showing but I'd rather just use a keyboard.
I miss the days when people thought keyloggers and spyware were a bad thing.
Highly agree!
So glad you are doing the 12 days of Christmas! Learning a ton!
Thank you guys so much for pushing towards AGI!
Super excited to be taking my writing to the next level within the free tier! No way, thank you so much for this amazing opportunity!
Canvas is a fantastic feature that will greatly enhance my work. I look forward to the addition of project creation and project knowledge capabilities, enabling ChatGPT to comprehend and continue work within specific projects, with dedicated chats for each. Additionally, it would be beneficial if ChatGPT could remember content from other project-related chats to avoid the need for copying and pasting when reaching message or context length limits. This would streamline workflows and improve efficiency. Thank you, OpenAI!
Kids in 2034 will nostalgically say they grew up with ChatGPT stories at bedtime...
What a sad vision.
@@thejeffyb9766 you're just old
@@thejeffyb9766 ok, boomer.
@@woodoyoywouldn't u chose your parents better?
@@thejeffyb9766 On the contrary, my kids love interacting with advanced voice mode to generate stories, prompting and tweaking as we go. Previously id get frustrated reading to them and having them interrupt, but the interactivity makes it enjoyable for all of us.
This is FANTASTIC, it may not be a function everyone uses, but for those how do this is REALLY good
Highlights:
- Launch of Canvas for all users, integrating it into the main ChatGPT model.
- Ability to run Python code directly within Canvas, providing instant feedback.
- New side-by-side view for easier collaboration and editing.
- Features include commenting on essays and debugging code with ChatGPT’s assistance.
- Custom GPTs can now leverage the power of Canvas for enhanced functionality.
I just want advanced voice mode to work with FILESSSSS
They‘ll surely update the advanced voice mode in the next couple of days, but I dont know, if they integrate file reading with it. I rather thought of the vision feature.
I want vision for advanced voice mode and singing, I want chatgpt to sing me a lullaby when I go to sleep
@@omegangerman I just want advanced voice mode to work with files cuz as a student. The advance voice mode really helps a lot cuz the conversations are just seamless. With the normal voice mode it takes like 6 second to wait for a response
What exactly do you mean?
Work with what files
That would be awesome! 🥰
Just a thought. I think OpenAI team members are really used to work with much better in-house version of these tools, they seemed genuinely confused and nervous when the public ChatGPT version just didn't do the task right at the first try during live demonstration 😁 The canvas tool has been probably already undergoing red teaming with O1 and who knows what else on board internally for the last few months at least.
As a developer, that's often the experience. You spend days preparing for a demo and everything works perfectly. Then, when it finally comes time for the real thing, it crashes, runs slowly, or doesn't run at all. It's frustrating and embarrassing and happens to just about everybody.
And still we clap to the idea that this time it would work at the first try
Hopefully, canvas will be in the desktop app.
ofcourse it is! the app is just a website wrapper. it's available right now i checked
you have to specifically say "open this in canvas". It took me forever to figure it out and thought it just wasn't possible on the desktop app
I've see now that Canvas is now available in the Windows desktop app. I've also noted that the OSx version can integrate into VS Code... I'm hoping that us Windows users get this integration feature soon.
@@zaubermanninc4390 Awesome! Thanks!
@@joshuapaul5171 Thanks so much for explaining this to me. 🙏
I have been using canvas for a while, and I think it is great. These improvements will help. However, it does become limited because it has a relatively short context. Once you get over a couple of hundred lines of code or maybe 1500 words, it becomes very unreliable. It will loose part of your edit or refuse to make changes and gas light you to say that it has made changes.
Can you please, please please increase the context length significantly of the canvas.
I had similar experiences. Unfortunately, this was not addressed in the video.
@@TillMiltzowthey’re gonna address memory windows or January 1 I’m pulling my funding
I have the same problem. I write blog posts that are over 2,000 words long and stories that are much longer. For shorter content, I've been very happy with canvas. For longer works, the feature is largely unusable at the moment.
Can you add HTML/CSS/JS preview in Canvas?
@tiisetsorammego3053 I ask about "run code"
with live preview
@@themastergronethis’d rock!
Dam, this feature I always wanted. Telling chatgpt what to update and seeing side by side how it will looks as a whole. This is amazing.
10:43, this has been happening to me so much where chat just freezes and I have to press the stop button. It’s so clunky, please fix this bug.
I don’t see it
You guys have changed my life and my future career. Thank you 🙏🙏
*Canvas Updates - Enhanced Collaboration and Functionality within ChatGPT*
* *0:12** Introduction:* Kevin Weil introduces Day 4 of "12 Days of OpenAI," focusing on Canvas updates.
* *0:31** Canvas Launch:* Canvas, previously in beta for Plus users, is now available for all users and integrated into the main ChatGPT model. It facilitates collaborative writing and coding beyond simple chat.
* *0:49** Key Features:* Three key updates are highlighted: 1) General availability of Canvas, 2) Python code execution within Canvas with visual output, and 3) Canvas integration with custom GPTs.
* *1:20** Collaborative Storytelling:* Lee demonstrates creating a Christmas story using Canvas, showcasing the side-by-side view for chat and document editing, collaborative editing features, and shortcut buttons for suggesting edits, adjusting length, refining reading level, polishing, and adding emojis.
* *5:25** Essay Feedback:* Alexi demonstrates using Canvas for essay feedback, highlighting the in-line commenting feature, which allows ChatGPT to provide targeted suggestions. The apply/dismiss buttons allow users to accept or reject proposed edits. The copy button facilitates easy transfer of the finalized text.
* *8:30** Programming Enhancements:* Canvas now supports Python code editing with syntax highlighting and autocomplete. A "Run" button enables code execution within Canvas, providing instant feedback.
* *9:13** Debugging with Canvas:* Alexi demonstrates debugging Python code in Canvas, showing how ChatGPT identifies errors, provides corrected code, and allows users to view changes.
* *11:17** Data Visualization:* Alexi demonstrates generating graphics (line chart and Sankey diagram) within Canvas by running Python code, showing how ChatGPT can assist with data visualization tasks.
* *13:38** Custom GPTs Integration:* Canvas is now integrated with custom GPTs, enabling users to incorporate Canvas functionality into their specialized GPTs.
* *14:00** Santa GPT Example:* Alexi demonstrates a custom GPT designed for Santa to draft letter responses, showcasing how Canvas can be automatically triggered based on custom instructions.
* *16:50** Custom GPT Configuration:* The process of configuring a custom GPT is shown, highlighting the instructions section where Canvas usage can be specified and the capabilities checkboxes for enabling Canvas functionality.
* *18:14** Rollout and Availability:* Canvas is rolling out to all web users, free and paid. Existing GPTs will need to manually enable Canvas through the configuration settings.
I used gemini-exp-1121 on rocketrecap dot com to summarize the transcript.
Cost (if I didn't use the free tier): $-0.0211
Input tokens: 20609
Output tokens: 522
Honestly, one of the better ideas for a feature.... is that when you start a new chat, chatGPT could use the context of this new conversation to perform some semantic similarity to previous conversations. The idea being that you may have had a conversation in the past that you have forgotten about..... but it relates heavily to what you are currently engaging chatgpt about. Or, that you know you spoke with chatGPT about it before, but instead of searching through the hundreds if not thousands of individual threads, chatgpt could just relate your new inquiry in some seamless fashion.... for example: "Oh yeah, I remember you talking about that back in September.... we spoke about x,y,z at length. Would you like a very brief summary of that thread?"
A smart search if you will that is chat aligned and natural feeling.
In this video you guys show that the feature is super buggy, it hangs, code does not work, when it does not throw errors the resulting pic does not make sense at all… the face of Kevin tell everything I need to know of the readiness of Canvas
Great news! This will help out tremendously with writing song lyrics.
Thank you for succinctly stating what exactly is new today within the first 60 seconds of the video. I hope later videos follow this trend.
Thank you so much OpenAI! This is a great feature!
Guys at Open ai are really
geniuses.Great going👍👍👍
I'm foreign English teacher in Taiwan being one of those like me, carries a lot of challenges, but hey, thanks for this awesome updates, teacher like me has a teaching assistant that gets the job done. Congratulations!
This is the best feature so far, great work 👍🏼
Out of all the things yet, I love canvas the most. Really useful for my work
PLEASE get rid of the insane amount of whitespace around Canvas so I can make the left side bigger.
You can resize it (at least in desktop version). There's a handle in between canvas and chat
@boevayasemga problem is, the right side readjusts to add even more white space. Shrinks the content area instead of filling it fully.
There's like 2 sizes, a smaller and a larger width. It snapping to the smaller width is what I have a problem with. Too much white space when it should just fully fill the area at those sizes.
@@PaulShellDev Agreed
Code execution is truly a game-changer. You get to see the results right before integrating the final piece of code.
I wish OpenAI made a Cursor competitor. There really is no substitution for having context within the IDE, and copy-pasting gets annoying for larger projects.
OpenAI backs Cursor so not sure this will happen.
Why not just use Cursor?
@@jeffbaird4318Because you have to pay $20 a month for OpenAI and $20 a month for cursor with limited 4o usage
Ketvirta diena tikrai įspūdinga! Bet būtų smagu, jei tekstų kiekis, su kuriais reikia bendrauti, būtų šiek tiek mažesnis - kartais per daug apkrauna. Vis dėlto, labai įdomu stebėti, kaip OpenAI juda į priekį!
Will these features come to CoPilot for VS code?
This was some fun banter with the Santa stuff. Made it very fun and not stiff.
Canvas needs more language support but this is great start with Python. Though would have preferred to start with HTML/CSS/JS & React like Claude Artifacts.
Let's add PowerShell to the mix as well.
Yeah, but how many projects we work on are a single file? I've got projects with multiple modules. This is where Anthropic gets it with their "Projects".
GhatGPT just feels like a neat toy for developing smaller custom tooling, as opposed to a tool to accomplish real-world work.
Just use Cursor then
Awesome work!!!
Oh, they added Claude's artifacts.
@@sdstorm improved*
Nah i think claude's artitifacts is better.@@youngop
@ you didn’t even use this new canvas yet, your mind is made up 😂
This has been there for a while for plus users. They've updated some features though.
artifacts is still better
The physics essay on the reindeers is hilarious! 😂
New canvas is amazing, any chance it could only update the code selection instead of rewriting it all ?
Any plans on integrating JavaScript/TypeScript to the Canvas?
IKR
this feature is the best one but WHY IS IT NOT ON O1 ???? we have had canvas for gpt4 for awhile
Please make an Open AI official VS Code Plugin, pleaaase (with the whole project context)
• o1 model - data analytics / interpretation (technical & scientfic field)
• Sora - visual rendering (creative industry like special visual effects)
• Canvas - text writing (can ChatGPT differentiate whether what it wrote is factual based or fictional creation ?)
When Canvas came out with pages of draft (be it for movie screenplay, film scriptwriting or best selling novel), would it be able to grade the content it produced will be a blockbuster success or a bestseller event when it gets 'published' out to the market (maybe to an 80% accuracy rate) ? → just an attempt to gauge its commercial viability upon its 'democratized' adoption out there.
Actually laughed out loud at Alexi's christmas list. Pretty sure we all want a h100. Nice work team OpenAI!
amazing features, great job guys
The think I like the most about OpenAI and their products is that they are thee most minimal with greatest user interface
Code editor is really nice. WIll Javascript be added?
Amazing new features! Congrats!
Most exciting release thus far!
Great, only I just finished my final seminar class and this would’ve definitely been useful. Still a great tool. Great job!!!
How about the following things (that I've seen popping up here):
- Enable advanced voice mode, even when interacting with uploaded files, graphics, or input text.
Then we could prompt how to interact with the model. Would this be possible?
- Integrate Whisper for voice interactions in the browser, similar to the desktop app's capabilities.
- Add folders or enhanced options for organizing chats, projects, and prompts.
- Allow the use of o1 or o1-mini in Custom GPTs for improved functionality.
Man, this is just day 4? I can´t imagine what´s coming next!
No canvas for o1? Ugh.
also no DALL E
Give it time. I'm sure we'll see it soon.
Be it online search or Canvas, with the new features that are being added, I still lack the function to deactivate them if necessary without having to write it down every time
You can turn it off in custom gpt settings
You guys are crushing it🔥
Great friendly presentation with nice Intelligent humour))
as long its ensured to be supportive , agi is much much awaited!!
Voice mode plus visual output please :)
When will the new canvas and projects features be released for the MacOS desktop app? @openai
This is amazing thank you for this gifts guys!
The Canvas preview has been a game changer for me - and being completely honest, is the main reason I kept my subscription with OpenAI as I've been very tempted by switching to the competition.
Python in canvas is interesting, but the real power will be integrating Canvas with users IDE of choice. At first I'm thinking "If only I could run .mjs in Canvas..." but in reality, if only Canvas was integrated directly with my VS Code node.js session.
I'm lead to believe that the OSx version of ChatGPT may do this, so I guess my question is when will we see this direct canvas integration feature for Windows/VS Code?
This will save many students' lives
Im always amazed about the frontend of chstgpt its so besutifull and the micro animations and ux are so good
Will Canvas in the future do javascript execution in some form? When will it come to the MacOS App?`
Didn’t we basically have this feature already lmao… I was thinking that shipmas for 12 days eventually it wouldn’t be very interesting… didn’t think it would be in the 4th day.
When you run out of interesting Christmas things to put on the shelf in the background, just grab a can of mountain dew to balance everything out.
Can you please increase the context window so we can upload abt 5000 lines of code and edit in the Canvas? That way we don’t have to write code in stages. Regular context size makes the system crash before completion.
Props to Alexi for working at OpenAI, dude graduated in 2020 from Yale, practically a new grad
Combining the power of code with tools like Rider and VS 2022 could revolutionize how we develop entire projects.
You guys are awesome 😎
I have found one additional issue with the letter to Santa. Canada Post is on strike, so Santa can't get his letters.
Will Canvas eventually be released for the desktop app?
When canvas will be released for all plus users?
keep up the great work 👍🏻 , fan of canvas cheers 🍻
Είμαι τόσο χαρούμενος, γιατί μου θυμίζετε τρεις παλιούς μου φίλους εδώ στη Mύκονο στην Eλλάδα.
Tον Bασίλη, τον Tσόπελα και τον Απόλλωνα... Τον καιρό του "Bolero Bar"
Day 11: AGI
Day 12: ASI
You've totally lost your mind. Day 11. sh*t, Day 12: another sh*t.
@@xevil21 Day 13: API
Day 14: ABI
Day 15: AMI (M for mind)
Day 69: BBC
@@somebody-anonymous Day 16: Close a.i. :D
Can you add more lenguages support (not only Python)??? Typescript, JS, Java, Kotlin, PHP, c, c++, etc.
It’s easier to automate python it has so many libraries really hard for other languages even more for C
@gamers-xh3uc And NodeJS?
This is awesome, exactly what I've been looking for
Wow running code inside canvas? That’s definitely a great Christmas present!
Fantastic, but the canvas in my GPT is uneditable and its tool button is greyed out. Help!!!
Canvas only becomes useful to programmers if the output token count is significantly increased (as in o1-mini)
Why would you use it instead of cursor or windsurf? It doesn't have any context of your codebase
This is actually the most useful feature for me 😮
I can't get into SORA either, if your preparation wasn't good enough for issues during a SORA launch, what what will happen when you push out AGI. "We’re currently experiencing heavy traffic and have temporarily disabled AGI safety protocols. If you've never filled out your will before, please fill it out soon."
It's about time GPTs got some type of update. This is excellent as an author.
When can we use Sora account creation isn't working
I guess we will have to wait next year to maybe see Canvas become multiplayer...
Nice update, quite useful. A new Version of DALL-E would be a banger!
What I didn’t understand is this only for 4o or for o1 as well
it's insane! love you oepn ai
Will it be possible to add 91 models to custom GPT’s?
Thank you for the really new feature : code execution. But could you give us a Canvas that produces and edits an HTML + JavaScript + CSS page with code execution ?
It’s the kind of single-page application that would be useful for everyone. (Python is just one option among other languages.) Editing an HTML page containing JavaScript shouldn’t be that difficult and seems equally universal in terms of usage.
when will canvas come to o1?
Will Csnvas run with Blender (open source with unique flavor of Python) or other 3D spp, or can I link to run the code via the built-in or another text editor?