Industry observers say GPT-4.5 is an “odd” model, question its price

OpenAI has introduced the discharge of GPT-4.5, which CEO Sam Altman beforehand stated can be the final non-chain-of-thought (CoT) mannequin. 

The corporate stated the brand new mannequin “shouldn’t be a frontier mannequin” however continues to be its largest giant language mannequin (LLM), with extra computational effectivity. Altman stated that, regardless that GPT-4.5 doesn’t cause the identical approach as OpenAI’s different new choices o1 or o3-mini, this new mannequin nonetheless presents extra human-like thoughtfulness. 

Business observers, a lot of whom had early entry to the brand new mannequin, have discovered GPT-4.5 to be an fascinating transfer from OpenAI, tempering their expectations of what the mannequin ought to have the ability to obtain. 

Wharton professor and AI commentator Ethan Mollick posted on social media that GPT-4.5 is a “very odd and fascinating mannequin,” noting it might probably get “oddly lazy on advanced tasks” regardless of being a robust author. 

OpenAI co-founder and former Tesla AI head Andrej Karpathy famous that GPT-4.5 made him bear in mind when GPT-4 got here out and he noticed the mannequin’s potential. In a post to X, Karpathy stated that, whereas utilizing GPT 4.5, “every thing is somewhat bit higher, and it’s superior, but in addition not precisely in methods which are trivial to level to.”

Karpathy, nevertheless warned that folks shouldn’t anticipate revolutionary influence from the mannequin because it “doesn’t push ahead mannequin functionality in circumstances the place reasoning is essential (math, code, and so on.).”

Business ideas intimately

Right here’s what Karpathy needed to say in regards to the newest GPT iteration in a prolonged put up on X:

“At the moment marks the discharge of GPT4.5 by OpenAI. I’ve been wanting ahead to this for ~2 years, ever since GPT4 was launched, as a result of this launch presents a qualitative measurement of the slope of enchancment you get out of scaling pretraining compute (i.e. merely coaching a much bigger mannequin). Every 0.5 within the model is roughly 10X pretraining compute. Now, recall that GPT1 barely generates coherent textual content. GPT2 was a confused toy. GPT2.5 was “skipped” straight into GPT3, which was much more fascinating. GPT3.5 crossed the brink the place it was sufficient to really ship as a product and sparked OpenAI’s “ChatGPT second”. And GPT4 in flip additionally felt higher, however I’ll say that it positively felt delicate.

I bear in mind being part of a hackathon looking for concrete prompts the place GPT4 outperformed 3.5. They positively existed, however clear and concrete “slam dunk” examples have been tough to seek out. It’s that … every thing was just a bit bit higher however in a diffuse approach. The phrase alternative was a bit extra artistic. Understanding of nuance within the immediate was improved. Analogies made a bit extra sense. The mannequin was somewhat bit funnier. World information and understanding was improved on the edges of uncommon domains. Hallucinations have been a bit much less frequent. The vibes have been only a bit higher. It felt just like the water that rises all boats, the place every thing will get barely improved by 20%. So it’s with that expectation that I went into testing GPT4.5, which I had entry to for a couple of days, and which noticed 10X extra pretraining compute than GPT4. And I really feel like, as soon as once more, I’m in the identical hackathon 2 years in the past. Every part is somewhat bit higher and it’s superior, but in addition not precisely in methods which are trivial to level to. Nonetheless, it’s unimaginable fascinating and thrilling as one other qualitative measurement of a sure slope of functionality that comes “totally free” from simply pretraining a much bigger mannequin.

Needless to say that GPT4.5 was solely educated with pretraining, supervised finetuning and RLHF, so this isn’t but a reasoning mannequin. Subsequently, this mannequin launch doesn’t push ahead mannequin functionality in circumstances the place reasoning is essential (math, code, and so on.). In these circumstances, coaching with RL and gaining considering is extremely vital and works higher, even whether it is on high of an older base mannequin (e.g. GPT4ish functionality or so). The cutting-edge right here stays the complete o1. Presumably, OpenAI will now be seeking to additional prepare with reinforcement studying on high of GPT4.5 to permit it to assume and push mannequin functionality in these domains.

HOWEVER. We do really anticipate to see an enchancment in duties that aren’t reasoning heavy, and I might say these are duties which are extra EQ (versus IQ) associated and bottlenecked by e.g. world information, creativity, analogy making, normal understanding, humor, and so on. So these are the duties that I used to be most occupied with throughout my vibe checks.

So beneath, I assumed it could be enjoyable to spotlight 5 humorous/amusing prompts that take a look at these capabilities, and to prepare them into an interactive “LM Area Lite” proper right here on X, utilizing a mix of photographs and polls in a thread. Sadly X doesn’t permit you to embrace each a picture and a ballot in a single put up, so I’ve to alternate posts that give the picture (exhibiting the immediate, and two responses one from 4 and one from 4.5), and the ballot, the place folks can vote which one is healthier. After 8 hours, I’ll reveal the identities of which mannequin is which. Let’s see what occurs 🙂“

Field CEO’s ideas on GPT-4.5

Different early customers additionally noticed potential in GPT-4.5. Field CEO Aaron Levie said on X that his firm used GPT-4.5 to assist extract structured knowledge and metadata from advanced enterprise content material. 

“The AI breakthroughs simply maintain coming. OpenAI simply introduced GPT-4.5, and we’ll be making it obtainable to Field prospects later at the moment within the Field AI Studio.

We’ve been testing GPT4.5 in early entry mode with Field AI for superior enterprise unstructured knowledge use-cases, and have seen robust outcomes. With the Field AI enterprise eval, we take a look at fashions towards a wide range of completely different situations, like Q&A accuracy, reasoning capabilities and extra. Specifically, to discover the capabilities of GPT-4.5, we centered on a key space with vital potential for enterprise influence: The extraction of structured knowledge, or metadata extraction, from advanced enterprise content material. 

At Field, we rigorously consider knowledge extraction fashions utilizing a number of enterprise-grade datasets. One key dataset we leverage is CUAD, which consists of over 510 industrial authorized contracts. Inside this dataset, Field has recognized 17,000 fields that may be extracted from unstructured content material and evaluated the mannequin primarily based on single shot extraction for these fields (that is our hardest take a look at, the place the mannequin solely has as soon as probability to extract all of the metadata in a single go vs. taking a number of makes an attempt). In our exams, GPT-4.5 accurately extracted 19 share factors extra fields precisely in comparison with GPT-4o, highlighting its improved skill to deal with nuanced contract knowledge.

Subsequent, to make sure GPT-4.5 may deal with the calls for of real-world enterprise content material, we evaluated its efficiency towards a extra rigorous set of paperwork, Field’s personal problem set. We chosen a subset of advanced authorized contracts – these with multi-modal content material, high-density data and lengths exceeding 200 pages – to symbolize among the most tough situations our prospects face. On this problem set, GPT-4.5 additionally persistently outperformed GPT-4o in extracting key fields with larger accuracy, demonstrating its superior skill to deal with intricate and nuanced authorized paperwork.

Total, we’re seeing robust outcomes with GPT-4.5 for advanced enterprise knowledge, which is able to unlock much more use-cases within the enterprise.“

Questions on worth and its significance

Whilst early customers discovered GPT-4.5 workable — albeit a bit lazy — they questioned its launch. 

As an illustration, distinguished OpenAI critic Gary Marcus referred to as GPT-4.5 a “nothingburger” on Bluesky.

Scorching take: GPT 4.5 is a nothingburger; GPT-5 nonetheless fantasy.• Scaling knowledge shouldn’t be a bodily regulation; just about every thing I instructed you was true.• All of the BS about GPT-5 we listened to for previous few years: not so true.• Fanboys like Cowen will blame customers, however outcomes simply aren’t what that they had hoped.

— Gary Marcus (@garymarcus.bsky.social) 2025-02-27T20:44:55.115Z

Hugging Face CEO Clement Delangue commented that GPT4.5’s closed-source provenance makes it “meh.” 

Nonetheless, many famous that GPT-4.5 had nothing to do with its efficiency. As an alternative, folks questioned why OpenAI would release a model so expensive that it’s nearly prohibitive to make use of but is not as powerful as its other models. 

One person commented on X: “So that you’re telling me GPT-4.5 is price greater than o1 but it doesn’t carry out as effectively on benchmarks…. Make it make sense.”

Different X users posited theories that the excessive token value might be to discourage opponents like DeepSeek “to distill the 4.5 mannequin.”

DeepSeek turned an enormous competitor towards OpenAI in January, with {industry} leaders discovering DeepSeek-R1 reasoning to be as succesful as OpenAI’s — however extra inexpensive.