# Introducing GPT-4o mini in the API **URL:** \ **Category:** Announcements\ **Tags:** gpt-4o-mini\ **Created:** [July 18, 2024, 5:27pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594 "2024-07-18T17:27:13Z")\ **Posts on this page:** 20\ **Page:** 1
**Author:** ![jeffsharris](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/jeffsharris/32/619093_2.png) [@jeffsharris](https://community.openai.com/u/jeffsharris)\ **Post date:** [July 18, 2024, 5:27pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/1 "2024-07-18T17:27:13Z")
We just officially launched [GPT-4o mini](https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence)—our new affordable and intelligent small model that’s significantly smarter, cheaper, and just as fast as GPT-3.5 Turbo. Some details on the new model: - **Intelligence:** GPT-4o mini outperforms GPT-3.5 Turbo in textual intelligence—scoring 82% on MMLU compared to 69.8%—and multimodal reasoning. - **Price:** GPT-4o mini is more than 60% cheaper than GPT-3.5 Turbo, priced at 15¢ per 1M input tokens and 60¢ per 1M output tokens (roughly the equivalent of 2500 pages in a standard book). - **Modalities:** GPT-4o mini currently supports text and vision capabilities, and we plan to add support for audio and video inputs and outputs in the future. - **Languages: GPT-4o mini has improved multilingual understanding over GPT-3.5 Turbo across a wide range of non-English languages.** Like GPT-4o, GPT-4o mini has a 128k context window and a knowledge cut-off date of October 2023. It also supports up to 16,384 [max\_tokens](https://platform.openai.com/docs/api-reference/chat/create#chat-create-max_tokens). We plan to launch fine-tuning for GPT-4o mini in the coming days. You can learn more about GPT-4o mini in our [announcement blog](https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence) and [API documentation](http://platform.openai.com/docs/models/gpt-4o-mini), or by testing the model in [Playground](https://platform.openai.com/playground/chat?models=gpt-4o-mini). Excited to hear what you think! ---
**Author:** ![PaulBellow](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/paulbellow/32/597962_2.png) [@PaulBellow](https://community.openai.com/u/PaulBellow)\ **Post date:** [July 18, 2024, 5:33pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/3 "2024-07-18T17:33:24Z")
Great news! Keep up the good work! ---
**Author:** ![VeitB](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/veitb/32/712987_2.png) [@VeitB](https://community.openai.com/u/VeitB)\ **Post date:** [July 18, 2024, 5:33pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/4 "2024-07-18T17:33:56Z")
That sounds like a big jump compared to GPT 3.5 Turbo! More context, more possibilities at a much lower price: Congrats to the team! Looking forward to test it! ---
**Author:** ![jim](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/jim/32/1976_2.png) [@jim](https://community.openai.com/u/jim)\ **Post date:** [July 18, 2024, 5:46pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/5 "2024-07-18T17:46:52Z")
Super excited about this…BUT first initial tests: function calling is not as accurate as GPT-3.5 Turbo (which already was problematic), doesn’t seem to be following my schema set in tools as much as GPT-4o does. I assume this was tested and deemed great, which makes me think perhaps there is a better way to write my tooling? Concrete, best practices for this newer model?? (which I would LOVE to work for me btw…) ---
**Author:** ![noumanjavaid](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/noumanjavaid/32/349208_2.png) [@noumanjavaid](https://community.openai.com/u/noumanjavaid)\ **Post date:** [July 18, 2024, 6:17pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/6 "2024-07-18T18:17:43Z")
I hope that this is the start of the near end for our GOAT GPT-4 Not an AI Expert guys but please fix the issues with GPT-4o the Context and the clarity points can be taken from `Gemini-1.5-pro` that model with 2M Token is a beast. But I would still like to use GPT-4 over anything. However, how long before it out performs OPENAI in all use case is just a matter of time really. # We need to catch up # God Speed @jeffsharris and Team OPENAI ---
**Author:** ![sps](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/sps/32/7736_2.png) [@sps](https://community.openai.com/u/sps)\ **Post date:** [July 18, 2024, 6:31pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/7 "2024-07-18T18:31:18Z")
Congratulations on the launch @jeffsharris This is great news! I really like the pricing, multi-modal nature, and the response times that I’m seeing. AI going 📈, token costs going 📉. ---
**Author:** ![tom18](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/tom18/32/231873_2.png) [@tom18](https://community.openai.com/u/tom18)\ **Post date:** [July 18, 2024, 6:39pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/8 "2024-07-18T18:39:22Z")
is it safe to say that 3.5 is going away and being phased out? I’m asking because one of my vendors has my plan mapped to 3.5, and I want an upgrade to 4o mini. ---
**Author:** ![merefield](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/merefield/32/14812_2.png) [@merefield](https://community.openai.com/u/merefield)\ **Post date:** [July 18, 2024, 6:59pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/11 "2024-07-18T18:59:07Z")
Trying to distract us from banging our tables for GPT-5 😉 Just kidding. This is a fantastic offering at this price, further expanding access to good language processing, excellent!! ---
**Author:** ![PaulBellow](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/paulbellow/32/597962_2.png) [@PaulBellow](https://community.openai.com/u/PaulBellow)\ **Post date:** [July 18, 2024, 7:00pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/12 "2024-07-18T19:00:51Z")
It took me literally about 5 seconds to switch out the models, so it shouldn’t be a problem switching and reaping the benefits. ---
**Author:** ![SomebodySysop](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/somebodysysop/32/17613_2.png) [@SomebodySysop](https://community.openai.com/u/SomebodySysop)\ **Post date:** [July 18, 2024, 7:27pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/13 "2024-07-18T19:27:06Z")
Maximum output token limit for gpt-4o-mini? ---
**Author:** ![ericlaycock44](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/ericlaycock44/32/425457_2.png) [@ericlaycock44](https://community.openai.com/u/ericlaycock44)\ **Post date:** [July 18, 2024, 7:30pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/14 "2024-07-18T19:30:05Z")
Amazing! @jeffsharris are there any plans to make gpt-4o-mini available in the finetuning API? Cheers * * * Edit: ah whoops looks like your announcement said as much. Can’t wait! ---
**Author:** ![PaulBellow](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/paulbellow/32/597962_2.png) [@PaulBellow](https://community.openai.com/u/PaulBellow)\ **Post date:** [July 18, 2024, 7:34pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/15 "2024-07-18T19:34:57Z")
In the blog post it mentions coming in a few days. 🙂 Welcome to the community! ---
**Author:** ![ericlaycock44](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/ericlaycock44/32/425457_2.png) [@ericlaycock44](https://community.openai.com/u/ericlaycock44)\ **Post date:** [July 18, 2024, 7:36pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/17 "2024-07-18T19:36:15Z")
Thanks PaulBellow! I am _such_ a fan of the OpenAI finetuning API. Much easier interface than some other API providers we could mention … 🛌 🪨 ---
**Author:** ![martinrobson](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/martinrobson/32/397924_2.png) [@martinrobson](https://community.openai.com/u/martinrobson)\ **Post date:** [July 18, 2024, 7:49pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/19 "2024-07-18T19:49:29Z")
I’d like a “mini me” Just less Dr. Evil… ---
**Author:** ![jamie10](https://avatars.discourse-cdn.com/v4/letter/j/d2c977/32.png) [@jamie10](https://community.openai.com/u/jamie10)\ **Post date:** [July 18, 2024, 7:59pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/20 "2024-07-18T19:59:16Z")
Ive noticed what i find is an issue, especially with using images in 4o mini , if i upload an image in normal 4o and ask a question, it uses qpprox 1000 input tokens and 200 output tokens, if i run the same task in 4o mini the total number of tokens is nearly 40000. Maybe its just me but i think something seens to be wrong. Anyone else have this issue? ---
**Author:** ![sat.tecnologias](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/sat.tecnologias/32/614524_2.png) [@sat.tecnologias](https://community.openai.com/u/sat.tecnologias)\ **Post date:** [July 18, 2024, 8:15pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/21 "2024-07-18T20:15:19Z")
I am currently leveraging GPT-4o mini in a Google Apps Script to enhance functionality within a spreadsheet. The script is used to summarize, correct grammar, improve spelling, translate text, and refine descriptions. Additionally, it provides detailed event and time information related to development activities and data analysis directly within the spreadsheet. By integrating GPT-4o mini, I have significantly improved the accuracy and efficiency of these tasks while reducing costs by over 60% compared to GPT-3.5 Turbo. This integration not only enhances productivity but also enriches the user experience, making complex data processing seamless and effective. I am grateful for the advanced capabilities of GPT-4o mini, which have been instrumental in achieving these improvements. Thank you to the OpenAI team for this powerful tool! ---
**Author:** ![eawestwrites](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/eawestwrites/32/129371_2.png) [@eawestwrites](https://community.openai.com/u/eawestwrites)\ **Post date:** [July 18, 2024, 9:45pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/22 "2024-07-18T21:45:36Z")
Can fine tuning please be available to everyone like 3.5 16k? Many of us never got GPT 4 fine tuning and it stopped all of my development of long form sample fine tuning. ---
**Author:** ![PaulBellow](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/paulbellow/32/597962_2.png) [@PaulBellow](https://community.openai.com/u/PaulBellow)\ **Post date:** [July 18, 2024, 9:53pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/23 "2024-07-18T21:53:54Z")
Hey, good to see you! Would love an update on your fine-tuning for fiction adventures if you ever have time. Hope you’re well! ---
**Author:** ![eawestwrites](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/eawestwrites/32/129371_2.png) [@eawestwrites](https://community.openai.com/u/eawestwrites)\ **Post date:** [July 18, 2024, 10:00pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/24 "2024-07-18T22:00:05Z")
The problem became everyone was stuck on 3.5 16k… we kept waiting for GPT 4 to open up… but it’s still “request” only. Not even turbo. I’m hoping with this lower cost model they will again open up fine tuning to everyone… We have new software we are debuting this summer that makes writing with AI seamless in-line. More like how Notion AI works, but allows the author to pick any model, including NSFW models. If mini 4o can be fine tune open for everyone, most of us will use that to fine tune and make outlines and components. 🙂 ---
**Author:** ![MikeFara](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/mikefara/32/400357_2.png) [@MikeFara](https://community.openai.com/u/MikeFara)\ **Post date:** [July 18, 2024, 10:13pm UTC](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594/25 "2024-07-18T22:13:07Z")
Is the API for assistants the same schema? I am getting internal server error, I figure this might blow something up or it is down. [Next page](https://community.openai.com/t/introducing-gpt-4o-mini-in-the-api/871594.md?page=2)