GPT-4o Mini First Impressions: Fast, Cheap, & Dang Good.

GPT-4o mini is OpenAI's latest small AI model, designed to offer improved performance and cost-efficiency. It outperforms other leading small models on reasoning tasks, scoring 82% on the MMLU benchmark. Priced at 15 cents per million input tokens and 60 cents per million output tokens, it's more than 60% cheaper than GPT-3.5 Turbo. With a 128,000 token context window and support for text and vision tasks, GPT-4o mini aims to provide developers and businesses with a more affordable option for high-volume, simple AI tasks while maintaining high-quality outputs.

▼ Link(s) From Today’s Video:

-------------------------------------------------

▼ Extra Links of Interest:

Let's work together!

Thanks for watching Matt Video Productions! I make all sorts of videos here on Youtube! Technology, Tutorials, and Reviews! Enjoy Your stay here, and subscribe!

All Suggestions, Thoughts And Comments Are Greatly Appreciated… Because I Actually Read Them.

0:00 Introduction to GPT-4.0 Mini
0:30 Discovery and Announcement of GPT-4.0 Mini
1:00 Key Features of GPT-4.0 Mini
2:30 Benchmarks and Comparisons
3:30 Updates on Other OpenAI Projects
4:30 First Impressions and Testing
7:00 Image Recognition Capabilities
10:00 Model Comparison and Self-awareness Test
11:30 Conclusion and Future Expectations
12:00 Closing Remarks

[Mesopotamia-vd]

We are getting GTA 6 before the voice mode

[TheMR-]

"Pineapple Pro" that was creative TBH.

[BunnyOfThunder]

Don't underestimate the importance of cheaper.

[henrytuttle]

A bullet fired horizontally would land at the same time. Vertically, the bullet dropped from the hand would hit the ground far sooner. Either the AI was using a canned response to the question, assumed you meant horizontally, or is just very bad.

[Dude_Wassup]

Anything but GPT 5. It’s like Rockstar with the constant updates of GTA 5, but no GTA 6. (Before they announced it.. finally.)

[CSlush]

I'm not sure how I feel about GPT-4o understanding that meme better than I did...

[dievas_]

People and businesses who are developing LLM apps needed this a lot more than any other dumb feature or another slow and expensive model.

[yhwhlungs]

GPT-4o models don't know they are in the omni family, they just know they are in the category of GPT-4.

[rodvik]

why would I want a model that "resists jailbreaks"? I want the model to do what I tell it, not be censored according to what OpenAi thinks is appropriate for me. I am not a child I am a paying adult.

[TheRealUsername]

15:01 These models aren't trained with their exact name, they are often referred to as "ChatGPT" based on "GPT-4" during fine-tuning and RLHF

[SuperFynn]

15:36 GPT4o and GPT4o Mini both don't know that they are 4o or 4o mini, both say when you ask them that they are based on the GPT4 architecture, but don't know anything about an Omni model or mini model (at least not when I asked)

[Saerthen]

You just invented the "mirror test" for LLMs.

[artemisfauls]

When will the voice mode be released, as it was in the presentation, do you have a prognosis?

[JC-jzrx]

11:42 damn, threw* a whole oak trees worth of shade at google with that one

[hipjoeroflmto]

I see all the comments from people who already pay for gpt being bored by this post but i see this as awesome cuz there replacing the free version of gpt

[Scn]

It looks like we can access 4o Mini in ChatGPT (besides API). Mini is not listed in the dropdown menu at the top left. However, if I start a conversation and then click the lightning bolt or star icon under the response, I'm able to choose "GPT-4o mini" from the dropdown

[davidyoung]

Non-numbered models in the OpenAI playground (or API) are always the latest version of that model, whereas numbered ones are "pinned".

[Felipe-zlrj]

9:57 the firearm shot vertically makes no sense at all. The model was wrong. It only would be right if it was horizontally.

[masterapofis]

They should have warned us that GPT-4o mini only has 50 questions every 4 hours at this rate.
Does it force you to have multiple accounts or have to share it, ceasing to be free, and when does that happen?
We will go to another AI. We want GPT-3.5 back!

[saksham]

Using the gpt-4o-mini API, when I ask, "Which model are you?" I get the reply, "I am based on OpenAI's GPT-3 model, which is designed to..." 🤔
Even in the playground, the same thing happens. However, when I select gpt-4o-mini in the Android app and ask the same question, it says it is based on the gpt-4 architecture.
This is very strange! 😱