LLM/AI Changelog

Creators can now choose from the full set of ChatGPT models (GPT-4o, o3, o4-mini and more) when building Custom GPTs—making it easier to fine-tune performance for different tasks, industries, and workflows. Creators can also set a recommended model to guide users.
Key details:
GPTswithout Custom Actionscan use the model picker to select from all models available to the user.GPTswith Custom Actionscurrently support GPT-4o and 4.1Available onwebfor users onPlus, Pro and Teamplans.Enterprise and Edu rollout coming soon.

Starting today, we’re adding several updates to projects in ChatGPT to help you do more focused work. These updates are available for Plus, Pro, and Team users.
Deep research and voice mode supportImprovements to memory to reference past chats in a project*Sharing chats from projectsStarting a new project directly from a chatUpload files and access model selector on mobile
Learn moreabout projects.
*Memory improvements are available for Plus and Pro users.

Like o1-pro, o3-pro is a version of our most intelligent model, o3, designed to think longer and provide the most reliable responses. Since the launch of o1-pro, users have favored this model for domains such as math, science, and coding—areas where o3-pro continues to excel, as shown in academic evaluations. Like o3, o3-pro has access to tools that make ChatGPT useful—it can search the web, analyze files, reason about visual inputs, use Python, personalize responses using memory, and more. Because o3-pro has access to tools, responses typically take longer than o1-pro to complete. We recommend using it for challenging questions where reliability matters more than speed, and waiting a few minutes is worth the tradeoff.
In expert evaluations, reviewers consistently prefer o3-pro over o3 in every tested category and especially in key domains like science, education, programming, business, and writing help. Reviewers also rated o3-pro consistently higher for clarity, comprehensiveness, instruction-following, and accuracy.
Academic evaluations show that o3-pro consistently outperforms both o1-pro and o3.
To assess the key strength of o3-pro, we once again use our rigorous "4/4 reliability" evaluation, where a model is considered successful only if it correctly answers a question in all four attempts, not just one:
o3-pro is available in the model picker for Pro and Team users starting today, replacing o1-pro. Enterprise and Edu users will get access the week after.
As o3-pro uses the same underlying model as o3, full safety details can be found in theo3 system card.
Limitations
At the moment, temporary chats are disabled for o3-pro as we resolve a technical issue.
Image generation is not supported within o3-pro—please use GPT-4o, OpenAI o3, or OpenAI o4-mini to generate images.
Canvas is also currently not supported within o3-pro.

Like o1-pro, o3-pro is a version of our most intelligent model, o3, designed to think longer and provide the most reliable responses. Since the launch of o1-pro, users have favored this model for domains such as math, science, and coding—areas where o3-pro continues to excel, as shown in academic evaluations. Like o3, o3-pro has access to tools that make ChatGPT useful—it can search the web, analyze files, reason about visual inputs, use Python, personalize responses using memory, and more. Because o3-pro has access to tools, responses typically take longer than o1-pro to complete. We recommend using it for challenging questions where reliability matters more than speed, and waiting a few minutes is worth the tradeoff.
In expert evaluations, reviewers consistently prefer o3-pro over o3 in every tested category and especially in key domains like science, education, programming, business, and writing help. Reviewers also rated o3-pro consistently higher for clarity, comprehensiveness, instruction-following, and accuracy.
Academic evaluations show that o3-pro consistently outperforms both o1-pro and o3.
To assess the key strength of o3-pro, we once again use our rigorous "4/4 reliability" evaluation, where a model is considered successful only if it correctly answers a question in all four attempts, not just one:
o3-pro is available in the model picker for Pro and Team users starting today, replacing o1-pro. Enterprise and Edu users will get access the week after.
As o3-pro uses the same underlying model as o3, full safety details can be found in theo3 system card.
Limitations
At the moment, temporary chats are disabled for o3-pro as we resolve a technical issue.
Image generation is not supported within o3-pro—please use GPT-4o, OpenAI o3, or OpenAI o4-mini to generate images.
Canvas is also currently not supported within o3-pro.

We're upgrading Advanced Voice in ChatGPT for paid users with significant enhancements in intonation and naturalness, making interactions feel more fluid and human-like. When we first launched Advanced Voice, it represented a leap forward in AI speech—now, it speaks even more naturally, with subtler intonation, realistic cadence (including pauses and emphases), and more on-point expressiveness for certain emotions including empathy, sarcasm, and more.Voice also now offers intuitive and effective language translation. Just ask Voice to translate between languages, and it will continue translating throughout your conversation until you tell it to stop or switch. It’s ready to translate whenever you need it—whether you're asking for directions in Italy or chatting with a colleague from the Tokyo office. For example, at a restaurant in Brazil, Voice can translate your English sentences into Portuguese, and the waiter’s Portuguese responses back into English—making conversations effortless, no matter where you are or who you're speaking with.This upgrade to Advanced Voice is available for all paid users across markets and platforms—just tap the Voice icon in the message composer to get started.This update is in addition to improvements we made earlier this year to ensure fewer interruptions and improved accents.
Known Limitations
In testing, we've observed that this update may occasionally cause minor decreases in audio quality, including unexpected variations in tone and pitch. These issues are more noticeable with certain voice options. We expect to improve audio consistency over time.Additionally, rare hallucinations in Voice Mode persist with this update, resulting in unintended sounds resembling ads, gibberish, or background music. We are actively investigating these issues and working toward a solution.

We're upgrading Advanced Voice in ChatGPT for paid users with significant enhancements in intonation and naturalness, making interactions feel more fluid and human-like. When we first launched Advanced Voice, it represented a leap forward in AI speech—now, it speaks even more naturally, with subtler intonation, realistic cadence (including pauses and emphases), and more on-point expressiveness for certain emotions including empathy, sarcasm, and more.
Voice also now offers intuitive and effective language translation. Just ask Voice to translate between languages, and it will continue translating throughout your conversation until you tell it to stop or switch. It’s ready to translate whenever you need it—whether you're asking for directions in Italy or chatting with a colleague from the Tokyo office. For example, at a restaurant in Brazil, Voice can translate your English sentences into Portuguese, and the waiter’s Portuguese responses back into English—making conversations effortless, no matter where you are or who you're speaking with.
This upgrade to Advanced Voice is available for all paid users across markets and platforms—just tap the Voice icon in the message composer to get started.
This update is in addition to improvements we made earlier this year to ensure fewer interruptions and improved accents.
Known Limitations
In testing, we've observed that this update may occasionally cause minor decreases in audio quality, including unexpected variations in tone and pitch. These issues are more noticeable with certain voice options. We expect to improve audio consistency over time.
Additionally, rare hallucinations in Voice Mode persist with this update, resulting in unintended sounds resembling ads, gibberish, or background music. We are actively investigating these issues and working toward a solution.

We're upgrading Advanced Voice in ChatGPT for paid users with significant enhancements in intonation and naturalness, making interactions feel more fluid and human-like. When we first launched Advanced Voice, it represented a leap forward in AI speech—now, it speaks even more naturally, with subtler intonation, realistic cadence (including pauses and emphases), and more on-point expressiveness for certain emotions including empathy, sarcasm, and more.
Voice also now offers intuitive and effective language translation. Just ask Voice to translate between languages, and it will continue translating throughout your conversation until you tell it to stop or switch. It’s ready to translate whenever you need it—whether you're asking for directions in Italy or chatting with a colleague from the Tokyo office. For example, at a restaurant in Brazil, Voice can translate your English sentences into Portuguese, and the waiter’s Portuguese responses back into English—making conversations effortless, no matter where you are or who you're speaking with.
This upgrade to Advanced Voice is available for all paid users across markets and platforms—just tap the Voice icon in the message composer to get started.
This update is in addition to improvements we made earlier this year to ensure fewer interruptions and improved accents.
Known Limitations
In testing, we've observed that this update may occasionally cause minor decreases in audio quality, including unexpected variations in tone and pitch. These issues are more noticeable with certain voice options. We expect to improve audio consistency over time.
Additionally, rare hallucinations in Voice Mode persist with this update, resulting in unintended sounds resembling ads, gibberish, or background music. We are actively investigating these issues and working toward a solution.

ChatGPT Team, Enterprise, and Edu customers globally can use connectors in deep research, as well as Pro and Plus users (excluding users in Switzerland, EEA, and the UK) to generate long-form, cited responses that include your company’s internal tools.
Supported connectors: Google Drive, SharePoint, Dropbox, Box, Outlook, Gmail, Google Calendar, Linear, GitHub, HubSpot, and TeamsCombines internal + web sources for synthesis
Learn more aboutConnectors in ChatGPT.

Admins and users can now build and deploy custom connectors to proprietary systems using Model Context Protocol (MCP).
Requires a remote MCP serverAvailable only in deep researchAdmin-published connectors appear in the connector list for all users
Learn more aboutbuilding custom connectors with MCP. For Team, Enterprise, and Edu plans, only admins can build and deploy custom connectors.

Memory improvements are starting to roll out to Free users. In addition to the saved memories that were there before, ChatGPT now references your recent conversations to deliver responses that feel more relevant and tailored to you.Free users must be logged in and on up-to-date apps (iOS/Android v1.2025.147+).
Opt‑in reminders
Free users in EEA (EU + UK), Switzerland, Norway, Iceland, or Liechtenstein will see a prompt to enable this setting or can visitSettings>Personalization>Memory>Reference chat historyto enable.Outside the European regions listed above, all Free users that have memory enabled will receive the upgrade automatically.
You can turn off memory anytime in settings. Learn more in ourMemory FAQ.

Memory improvements are starting to roll out to Free users. In addition to the saved memories that were there before, ChatGPT now references your recent conversations to deliver responses that feel more relevant and tailored to you.
Free users must be logged in and on up-to-date apps (iOS/Android v1.2025.147+).
Opt‑in reminders
Free users in EEA (EU + UK), Switzerland, Norway, Iceland, or Liechtenstein will see a prompt to enable this setting or can visitSettings>Personalization>Memory>Reference chat historyto enable.Outside the European regions listed above, all Free users that have memory enabled will receive the upgrade automatically.
You can turn off memory anytime in settings. Learn more in ourMemory FAQ.

Since its launch in the API in April, GPT-4.1 has become a favorite among developers—by popular demand, we’re making it available directly in ChatGPT.GPT-4.1 is a specialized model that excels at coding tasks. Compared to GPT-4o, it's even stronger at precise instruction following and web development tasks, and offers an alternative to OpenAI o3 and OpenAI o4-mini for simpler, everyday coding needs.Starting today, Plus, Pro, and Team users can access GPT-4.1 via the "more models" dropdown in the model picker. Enterprise and Edu users will get access in the coming weeks. GPT-4.1 has the same rate limits as GPT-4o for paid users.

GPT-4.1 mini is a fast, capable, and efficient small model, delivering significant improvements compared to GPT-4o mini—in instruction-following, coding, and overall intelligence. Starting today, GPT-4.1 mini replaces GPT-4o mini in the model picker under "more models" for paid users, and will serve as the fallback model for free users once they reach their GPT-4o usage limits. Rate limits remain the same.Evals for GPT-4.1 and GPT-4.1 mini were originally shared in theblog postaccompanying their API release. They also went through standard safety evaluations. Detailed results are available in the newly launchedSafety Evaluations Hub.

Since its launch in the API in April, GPT-4.1 has become a favorite among developers—by popular demand, we’re making it available directly in ChatGPT.
GPT-4.1 is a specialized model that excels at coding tasks. Compared to GPT-4o, it's even stronger at precise instruction following and web development tasks, and offers an alternative to OpenAI o3 and OpenAI o4-mini for simpler, everyday coding needs.
Starting today, Plus, Pro, and Team users can access GPT-4.1 via the "more models" dropdown in the model picker. Enterprise and Edu users will get access in the coming weeks. GPT-4.1 has the same rate limits as GPT-4o for paid users.

GPT-4.1 mini is a fast, capable, and efficient small model, delivering significant improvements compared to GPT-4o mini—in instruction-following, coding, and overall intelligence. Starting today, GPT-4.1 mini replaces GPT-4o mini in the model picker under "more models" for paid users, and will serve as the fallback model for free users once they reach their GPT-4o usage limits. Rate limits remain the same.
Evals for GPT-4.1 and GPT-4.1 mini were originally shared in theblog postaccompanying their API release. They also went through standard safety evaluations. Detailed results are available in the newly launchedSafety Evaluations Hub.

Since its launch in the API in April, GPT-4.1 has become a favorite among developers—by popular demand, we’re making it available directly in ChatGPT.
GPT-4.1 is a specialized model that excels at coding tasks. Compared to GPT-4o, it's even stronger at precise instruction following and web development tasks, and offers an alternative to OpenAI o3 and OpenAI o4-mini for simpler, everyday coding needs.
Starting today, Plus, Pro, and Team users can access GPT-4.1 via the "more models" dropdown in the model picker. Enterprise and Edu users will get access in the coming weeks. GPT-4.1 has the same rate limits as GPT-4o for paid users.

GPT-4.1 mini is a fast, capable, and efficient small model, delivering significant improvements compared to GPT-4o mini—in instruction-following, coding, and overall intelligence. Starting today, GPT-4.1 mini replaces GPT-4o mini in the model picker under "more models" for paid users, and will serve as the fallback model for free users once they reach their GPT-4o usage limits. Rate limits remain the same.
Evals for GPT-4.1 and GPT-4.1 mini were originally shared in theblog postaccompanying their API release. They also went through standard safety evaluations. Detailed results are available in the newly launchedSafety Evaluations Hub.

ChatGPT deep research with SharePoint and OneDrive is available globally to Team users. It is also gradually rolling out to Plus and Pro users, except for those in the EEA, Switzerland, and the UK. Enterprise user access will be announced at a later date.See:Connecting SharePoint and Microsoft OneDrive to ChatGPT deep research

ChatGPT deep research with Sharepoint and OneDrive is available globally to Team users. It is also gradually rolling out to Plus and Pro users, except for those in the EEA, Switzerland, and the UK. Enterprise user access will be announced at a later date.
See:Connecting SharePoint and Microsoft OneDrive to ChatGPT deep research

Enhanced memory rolling out to all Plus and Pro users (including the EU). The new memory features are available in the EEA (EU + UK), Switzerland, Norway, Iceland, or Liechtenstein. These features areOFF by defaultand must be enabled inSettings > Personalization > Reference  Chat  History.
Plan differences
•Saved  memoriesandChat historyare offered only to Plus and Pro accounts.• Free‑tier users have access toSaved  memoriesonly.
Opt‑in reminders
• Outside the European regions listed above, all Plus and Pro accounts that have memory enabled will receive the upgrade automatically.• If you previously opted out of memory, ChatGPT will not reference past conversations unless you opt back in.SeeMemory FAQ.

Enhanced memory rolling out to all Plus and Pro users (including the EU). The new memory features are available in the EEA (EU + UK), Switzerland, Norway, Iceland, or Liechtenstein. These features areOFF by defaultand must be enabled inSettings > Personalization > Reference  Chat  History.
Plan differences
•Saved  memoriesandChat historyare offered only to Plus and Pro accounts.
• Free‑tier users have access toSaved  memoriesonly.
Opt‑in reminders
• Outside the European regions listed above, all Plus and Pro accounts that have memory enabled will receive the upgrade automatically.
• If you previously opted out of memory, ChatGPT will not reference past conversations unless you opt back in.
SeeMemory FAQ.

We've removed the row of individual tool icons from the mobile composer and replaced it with the new sliders‑style icon to open theSkillsmenu; tapping that button opens a bottom‑sheet menu where users can choose tools like Create an image or Search the web.
No tools are deprecated—access is simply consolidated to clear space and reduce on‑screen clutter.

We've removed the row of individual tool icons from the mobile composer and replaces it with the new sliders‑style icon to open theSkillsmenu; tapping that button opens a bottom‑sheet menu where users can choose tools like Create an image or Search the web.
No tools are deprecated—access is simply consolidated to clear space and reduce on‑screen clutter.

We've reverted the most recent update to GPT-4o due to issues with overly agreeable responses (sycophancy).We’re actively working on further improvements. For more details, check out ourblog postexplaining what happened and our initial findings, andthis blog postwhere we expand on what we missed with sycophancy and the changes we're going to make going forward.

We've reverted the most recent update to GPT-4o due to issues with overly agreeable responses (sycophancy).
We’re actively working on further improvements. For more details, check out ourblog postexplaining what happened and our initial findings, andthis blog postwhere we expand on what we missed with sycophancy and the changes we're going to make going forward.

We've reverted the most recent update to GPT-4o due to issues with overly agreeable responses (sycophancy).
We’re actively working on further improvements. For more details, check out ourblog postexplaining what happened and our initial findings, andthis blog postwhere we expand on what we missed with sycophancy and the changes we're going to make going forward.

We’re making additional improvements to GPT-4o, optimizing when it saves memories and enhancing problem-solving capabilities for STEM. We’ve also made subtle changes to the way it responds, making it more proactive and better at guiding conversations toward productive outcomes. We think these updates help GPT-4o feel more intuitive and effective across a variety of tasks–we hope you agree!

We’re making additional improvements to GPT-4o, optimizing when it saves memories and enhancing problem-solving capabilities for STEM. We’ve also made subtle changes to the way it responds, making it more proactive and better at guiding conversations toward productive outcomes. We think these updates help GPT-4o feel more intuitive and effective across a variety of tasks–we hope you agree!

OpenAI o3is our most powerful reasoning model that pushes the frontier acrosscoding, math, science, visual perception, and more. It sets a new SOTA on benchmarks including Codeforces, SWE-bench (without building a custom model-specific scaffold), and MMMU. It’s ideal for complex queries requiring multi-faceted analysis and whose answers may not be immediately obvious. It performs especially strongly at visual tasks like analyzing images, charts, and graphics. In evaluations by external experts, o3 makes 20 percent fewer major errors than OpenAI o1 on difficult, real-world tasks—especially excelling in areas like programming, business/consulting, and creative ideation. Early testers highlighted its analytical rigor as a thought partner and emphasized its ability to generate and critically evaluate novel hypotheses—particularly within biology, math, and engineering contexts.
OpenAI o4-miniis a smaller model optimized for fast, cost-efficient reasoning—it achieves remarkable performance for its size and cost, particularly inmath, coding, and visual tasks. It is the best-performing benchmarked model on AIME 2024 and 2025. In expert evaluations, it also outperforms its predecessor, o3‑mini, on non-STEM tasks as well as domains like data science. Thanks to its efficiency, o4-mini supports significantly higher usage limits than o3, making it a strong high-volume, high-throughput option for questions that benefit from reasoning.

Today, we’re releasing OpenAIo3ando4-mini,the latest in our o-series of models trained to think for longer before responding. These are the smartest models we’ve released to date, representing a step change in ChatGPT's capabilities for everyone from curious users to advanced researchers. For the first time, our reasoning models can agentically use and combine every tool within ChatGPT—this includes searching the web, analyzing uploaded files and other data with Python, reasoning deeply about visual inputs, and even generating images. Critically, these models are trained to reason about when and how to use tools to produce detailed and thoughtful answers in the right output formats, typically in under a minute, to solve more complex problems. This allows them to tackle multi-faceted questions more effectively, a step toward a more agentic ChatGPT that can independently execute tasks on your behalf. The combined power of state-of-the-art reasoning with full tool access translates into significantly stronger performance across academic benchmarks and real-world tasks, setting a new standard in both intelligence and usefulness.

All images you create with ChatGPT are now automatically saved to a newLibraryin the sidebar, giving you one place to browse, revisit, and reuse your work without digging through past conversations. The Library is rolling out today on Web, iOS, and Android for Free, Plus, and Pro users (Enterprise / Edu support coming soon). For now, it displays images generated with 4o Image Generation while we backfill older creations, and you can remove an image by deleting the conversation where it was made.

Sunsetting GPT‑4 in ChatGPT
Effective  April  30,  2025, GPT‑4 will be retired from ChatGPT and fully replaced by GPT‑4o.GPT‑4o is our newer, natively multimodal model. In head‑to‑head evaluations it consistently surpasses GPT‑4 in writing, coding, STEM, and more.Recent upgrades have further improved GPT‑4o’s instruction following, problem solving, and conversational flow, making it a natural successor to GPT‑4.GPT-4 will still be available in the API.GPT‑4 marked a pivotal moment in ChatGPT’s evolution. We’re grateful for the breakthroughs it enabled and for the feedback that helped shape its successor. GPT‑4o builds on that foundation to deliver even greater capability, consistency, and creativity.

Sunsetting GPT‑4 in ChatGPT
Effective  April  30,  2025, GPT‑4 will be retired from ChatGPT and fully replaced by GPT‑4o.
GPT‑4o is our newer, natively multimodal model. In head‑to‑head evaluations it consistently surpasses GPT‑4 in writing, coding, STEM, and more.
Recent upgrades have further improved GPT‑4o’s instruction following, problem solving, and conversational flow, making it a natural successor to GPT‑4.
GPT-4 will still be available in the API.
GPT‑4 marked a pivotal moment in ChatGPT’s evolution. We’re grateful for the breakthroughs it enabled and for the feedback that helped shape its successor. GPT‑4o builds on that foundation to deliver even greater capability, consistency, and creativity.

We’ve made improvements to GPT-4o—it now feels more intuitive, creative, and collaborative, with enhanced instruction-following, smarter coding capabilities, and a clearer communication style.
Smarter problem-solving in STEM and coding:
GPT-4o has further improved its capability to tackle complex technical and coding problems. It now generates cleaner, simpler frontend code, more accurately thinks through existing code to identify necessary changes, and consistently produces coding outputs that successfully compile and run, streamlining your coding workflows.
Enhanced instruction-following and formatting accuracy:
GPT-4o is now more adept at following detailed instructions, especially for prompts containing multiple or complex requests. It improves on generating outputs according to the format requested and achieves higher accuracy in classification tasks.
“Fuzzy” improvements:
Early testers say that the model seems to better understand the implied intent behind their prompts, especially when it comes to creative and collaborative tasks. It’s also slightly more concise and clear, using fewer markdown hierarchies and emojis for responses that are easier to read, less cluttered, and more focused. We're curious to see if our users also find this to be the case.
This model is now available in ChatGPT and in the API as the newest snapshot of chatgpt-4o-latest. We plan to bring these improvements to a dated model in the API in the coming weeks.

We’ve made improvements to GPT-4o—it now feels more intuitive, creative, and collaborative, with enhanced instruction-following, smarter coding capabilities, and a clearer communication style.
Smarter problem-solving in STEM and coding:
GPT-4o has further improved its capability to tackle complex technical and coding problems. It now generates cleaner, simpler frontend code, more accurately thinks through existing code to identify necessary changes, and consistently produces coding outputs that successfully compile and run, streamlining your coding workflows.
Enhanced instruction-following and formatting accuracy:
GPT-4o is now more adept at following detailed instructions, especially for prompts containing multiple or complex requests. It improves on generating outputs according to the format requested and achieves higher accuracy in classification tasks.
“Fuzzy” improvements:
Early testers say that the model seems to better understand the implied intent behind their prompts, especially when it comes to creative and collaborative tasks. It’s also slightly more concise and clear, using fewer markdown hierarchies and emojis for responses that are easier to read, less cluttered, and more focused. We're curious to see if our users also find this to be the case.
This model is now available in ChatGPT and in the API as the newest snapshot of chatgpt-4o-latest. We plan to bring these improvements to a dated model in the API in the coming weeks.

1⁠. In-line message error retries: If you encounter a message error, you can retry or just continue chatting in the same conversation.
2⁠. o1 and o3-mini now offer Python-powered data analysis in ChatGPT: You can now ask these models to perform tasks like running regressions on test data, visualizing complex business metrics, and conducting scenario-based simulations.3⁠. Conversation Drafts: Unsubmitted messages in your message prompt will now be saved.
4⁠. New UI for Temporary Chats: We’ve made accessing and viewing these private conversations more clear.

1. In-line message error retries: If you encounter a message error, you can retry or just continue chatting in the same conversation.
2. o1 and o3-mini now offer Python-powered data analysis in ChatGPT: You can now ask these models to perform tasks like running regressions on test data, visualizing complex business metrics, and conducting scenario-based simulations.
3. Conversation Drafts: Unsubmitted messages in your message prompt will now be saved.
4. New UI for Temporary Chats: We’ve made accessing and viewing these private conversations more clear.