GPT-4o is the next big update to GPT-4, a new iteration on top of GPT-4 and GPT-4 Turbo. It's free for everyone, faster, cheaper to operate, and gives better results than ever before. OpenAI's CTO Mira Murati unveiled it at the company's Spring Update on May 13th, and thanks to how fast it is, you'll even be able to converse with GPT-4o out loud using your microphone.
The 'o' in GPT-4o stands for "omni", and its big focus is on multi-modal inputs. It can handle voice, video, images, and significantly improves non-English language support as well.
ChatGPT vs Microsoft Copilot vs Google Gemini: What are the differences?
If you've been trying to figure out which generative AI tool is better, you've come to the right place
GPT-4o vs GPT-4 and GPT-4 Turbo
Big changes for the average user
GPT-4o brings GPT-4 level intelligence to everyone, including free users. That means free users will finally get an upgrade over the free GPT-3.5, and that will be rolled out iteratively over the next few weeks.
This is OpenAI's new flagship model, meaning that it's the best GPT model from the company yet. It's much faster, improves on all of its capabilities, and can be used for basically anything. GPT-4 Turbo, previously, was only accessible either via OpenAI's API for it, or through Microsoft Copilot, where users could flip a switch to enable it.
On context windows, GPT-4o benefits from a much larger context window than GPT-4 and GPT-3.5, too. For reference, GPT-4 Turbo's context window is 128K tokens, with a maximum response token length of 4,096. Early reports suggest that the output token limit with GPT-4o has been halved to 2048, but OpenAI does say that it has a context window of 128K tokens and a cut-off date of October 2023.
Context windows in AI are windows that compromise the entirety of the AI's "thinking." This includes all the input it uses to understand and generate a response, encompassing the prompt given by the user as well as any additional context or previous dialogue. The size of the context window determines how much information the model can "remember" from earlier in the conversation or document, affecting its ability to produce coherent and relevant responses.
OpenAI mentioned that GPT-4o will be coming to the API, and there they mentioned some of the biggest improvements in terms of numbers. GPT-4o is two times faster, 50% cheaper, and has five times the rate limits when compared to GPT-4 Turbo. Those are big improvements, and GPT-4 Turbo was already significantly faster. Given the cost reduction in operating this model, it makes sense that OpenAI will also bring it out to free users.
The biggest improvement comes from its multi-modal use, though. It incorporates vision and voice, and can actively converse with a user. OpenAI demonstrated how you could speak with ChatGPT, and how it could identify emotion, help solve math equations, and so much more. The company didn't shy away from showing some of its limitations in the presentation either, but those were few and far between. OpenAI also says that GPT-4o can respond over voice in as little as 232 milliseconds, with an average of 320 milliseconds, and that it will be available to ChatGPT Plus subscribers in just a few weeks.
OpenAI says that it matches the performance of GPT-4 Turbo while being significantly faster while improving massively in non-English languages. Those languages include Korean, Russian, Chinese, Arabic, and more.
What is GPT-4 Turbo? The successor to GPT-4 that's available in preview now
GPT-4 Turbo is the successor to GPT-4, and this is everything you need to know about it.
How to use GPT-4o
OpenAI has said it's coming over the next few weeks
If you're looking to try out GPT-4o, you'll need to wait a while. All users will get it over the next few weeks with the rollout starting today, and paying customers will still get all of their other advantages like memory, custom GPTs, and more. OpenAI didn't give any specific dates or deadlines, presumably meaning that it's a staged rollout, possibly by region. Specifically, GPT-4o will be available in ChatGPT Free, Plus, and Team (with Enterprise coming soon) and in the Chat Completions API, Assistants API, and Batch API.
Free users will have a limit on how much they can use GPT-4o, and at times of peak demand, will default back to GPT-3.5. Free users will also be limited in how they can use advanced tools, such as:
- Data analysis
- File Uploads
- Browse
- Discovering and using GPTs
- Vision
OpenAI also says that Plus users will be able to send up to 80 messages every three hours on GPT-4o and up to 40 messages every three hours on GPT-4. This may be reduced in future if the company struggles to keep up with demand.
If you're a developer, you can already use GPT-4o, and a trusted set of partner developers will also be able to trial its audio and visual uses over an API in the coming weeks. You can keep checking your ChatGPT account to see if you have access.
ChatGPT
ChatGPT is one of the best LLMs out there, and it's completely free to use. You'll need to make an account, but there's not much else to it.