16. april 2025.

Introducing OpenAI o3 and o4-mini

Učitavanje…

Ažuriranje 10. juna 2025.: OpenAI o3‑pro je sada dostupan Pro korisnicima u ChatGPT‑u, kao i u našem API-ju. Kao i OpenAI o1‑pro, o3‑pro je verzija našeg najinteligentnijeg modela, OpenAI o3, dizajniranog da razmišlja duže i pruža najpouzdanije odgovore. Sve detalje možete pronaći u našim bilješkama o izdanju⁠(otvara se u novom prozoru).

Danas predstavljamo OpenAI o3 i o4-mini, najnovije u našoj o-seriji modela obučenih da razmišljaju duže prije nego što odgovore. Ovo su najpametniji modeli koje smo do sada objavili, predstavljajući značajan napredak u mogućnostima ChatGPT‑a za sve, od znatiželjnih korisnika do naprednih istraživača. Po prvi put, naši modeli rezonovanja mogu agentički koristiti i kombinovati svaki alat unutar ChatGPT‑a—ovo uključuje pretraživanje weba, analizu otpremljenih datoteka i drugih podataka pomoću Pythona, duboko rezonovanje o vizuelnim unosima, pa čak i generirati slike. Ključno je da su ovi modeli obučeni da rezoniraju o tome kada i kako koristiti alate za izradu detaljnih i promišljenih odgovora u pravim formatima izlaza, obično za manje od minute, kako bi riješili složenije probleme. Ovo im omogućava da efikasnije rješavaju složena pitanja, što je korak ka agentičnijem ChatGPT‑u koji može samostalno izvršavati zadatke u vaše ime. Kombinovana snaga najsavremenijeg rezonovanja sa potpunim pristupom alatima rezultira znatno jačim performansama u akademskim mjerilima i stvarnim zadacima, postavljajući novi standard u inteligenciji i korisnosti.

Šta se promijenilo

OpenAI o3 je naš najmoćniji model rezonovanja koji pomiče granice u kodiranju, matematici, nauci, vizualnoj percepciji i još mnogo čemu. Postavlja novi SOTA na benchmarkovima, uključujući Codeforces, SWE-bench (bez izgradnje prilagođenog scaffolda specifičnog za model) i MMMU. Idealan je za složene upite koji zahtijevaju višestruku analizu i čiji odgovori možda nisu odmah očigledni. Posebno snažno se ističe u vizualnim zadacima kao što su analiza slika, grafikona i grafike. U evaluacijama eksternih stručnjaka, o3 pravi 20 posto manje velikih grešaka nego OpenAI o1 na zahtjevnim, stvarnim zadacima — posebno se ističe u oblastima kao što su programiranje, poslovanje i kreativno osmišljavanje ideja. Rani testeri su istakli njegovu analitičku rigoroznost kao partnera za razmišljanje i naglasili njegovu sposobnost generiranja i kritičke evaluacije novih hipoteza — posebno u kontekstima biologije, matematike i inženjerstva.

OpenAI o4-mini je manji model optimiziran za brzo i isplativo rezonovanje—postiže izvanredne performanse za svoju veličinu i cijenu, posebno u matematici, kodiranju i vizuelnim zadacima. To je model s najboljim performansama prema AIME 2024 i 2025. Iako pristup računaru značajno smanjuje težinu AIME ispita, također smo primijetili da o4-mini postiže 99,5% prolaznost@1 (100% saglasnost@8) na AIME 2025 kada ima pristup Python interpreteru. Iako se ovi rezultati ne bi trebali upoređivati s performansama modela bez pristupa alatima, oni su primjer koliko efikasno o4-mini koristi dostupne alate; o3 pokazuje slična poboljšanja na AIME 2025 zahvaljujući upotrebi alata (98,4% pass@1, 100% consensus@8).

U stručnim evaluacijama, o4-mini također nadmašuje svog prethodnika, o3‑mini, u zadacima koji nisu STEM, kao i u domenama kao što je nauka o podacima. Zahvaljujući svojoj efikasnosti, o4-mini podržava znatno veće granice korištenja od o3, što ga čini snažnom opcijom za veliki obim i visok protok za pitanja koja imaju koristi od rezonovanja. Vanjski stručni evaluatori ocijenili su oba modela kao one koji pokazuju poboljšano praćenje instrukcija i korisnije, provjerljive odgovore od svojih prethodnika, zahvaljujući poboljšanoj inteligenciji i uključivanju web izvora. U poređenju s prethodnim iteracijama naših modela rezonovanja, ova dva modela trebala bi djelovati prirodnije i razgovornije, posebno jer se oslanjaju na memoriju i prethodne razgovore kako bi odgovori bili personalizovaniji i relevantniji.

Višemodalni

Kodiranje

Svi SWE-bench evaluacijski pokusi koriste fiksni podskup od n=477 verificiranih zadataka koji su potvrđeni na našoj internoj infrastrukturi.

Praćenje uputa i korištenje agentnih alata

Svi modeli se procjenjuju pri visokim postavkama „napora rezonovanja“ - slično varijantama poput 'o4-mini-high' u ChatGPT‑u.

Nastavak skaliranja učenja putem pojačanja

Tokom razvoja OpenAI o3, primijetili smo da pojačano učenje velikih razmjera pokazuje isti trend „više računanja = bolje performanse“ koji je uočen u pretreniravanju GPT‑serije. Ponavljanjem puta skaliranja—ovaj put u RL-u—pomaknuli smo se za dodatni red veličine u računanju za obuku i rezonovanje u vremenu inferencije, ali i dalje vidimo jasna poboljšanja performansi, potvrđujući da se performanse modela nastavljaju poboljšavati što im se više dozvoli razmisliti. Pri istoj latenciji i troškovima kao OpenAI o1, o3 pruža bolje performanse u ChatGPT‑u—i potvrdili smo da, ako mu dopustimo da duže razmišlja, njegove performanse nastavljaju rasti.

Također smo obučili oba modela da koriste alate kroz pojačano učenje—učeći ih ne samo kako da koriste alate, već i da rezoniraju o tome kada ih koristiti. Njihova sposobnost primjene alata na osnovu željenih ishoda čini ih sposobnijima u otvorenim situacijama—posebno onima koje uključuju vizualno rezonovanje i višestepene tokove rada. Ovo poboljšanje se ogleda i u akademskim mjerilima i u zadacima iz stvarnog svijeta, kako su izvijestili rani testeri.

Razmišljanje uz slike

Po prvi put, ovi modeli mogu direktno integrirati slike u svoj lanac misli. Oni ne vide samo sliku—već razmišljaju o njoj. Ovo otključava novu klasu rješavanja problema koja spaja vizualno i tekstualno rezonovanje, što se odražava u njihovim vrhunskim performansama na višemodalnim testovima.

Osobe mogu otpremiti fotografiju bijele ploče, dijagram iz udžbenika ili ručno nacrtanu skicu, a model ih može interpretirati - čak i ako je slika mutna, obrnuta ili niske kvalitete. Korištenjem alata, modeli mogu manipulirati slikama u hodu—rotirajući, zumirajući ili transformirajući ih kao dio procesa rezonovanja.

Ovi modeli pružaju najbolju tačnost u klasi za zadatke vizualne percepcije, omogućavajući im da riješe pitanja koja su ranije bila nedostižna. Posjetite blog o istraživanju vizualnog rezonovanja⁠ kako biste saznali više.

Prema upotrebi agentičkog alata

OpenAI o3 i o4-mini imaju potpuni pristup alatima unutar ChatGPT‑a, kao i vašim vlastitim prilagođenim alatima putem pozivanja funkcija u API-ju. Ovi modeli su obučeni da rezoniraju o tome kako riješiti probleme, birajući kada i kako koristiti alate za generisanje detaljnih i promišljenih odgovora u pravim formatima izlaza brzo—obično za manje od minute.

Na primjer, korisnik bi mogao pitati: „Kako će se ljetna potrošnja energije u Kaliforniji usporediti s prošlogodišnjom?“ Model može pretraživati web za podatke o javnim uslugama, pisati Python kod za izradu prognoze, generirati grafikon ili sliku i objasniti ključne faktore iza predviđanja, povezujući više poziva alata. Rezonovanje omogućava modelima da reaguju i prilagođavaju se prema potrebi informacijama na koje nailaze. Na primjer, mogu pretraživati web više puta uz pomoć pretraživača, pregledavati rezultate i isprobavati nove pretrage ako im treba više informacija.

Ovaj fleksibilan, strateški pristup omogućava modelima da se bave zadacima koji zahtijevaju pristup najnovijim informacijama izvan ugrađenog znanja modela, proširenog rezonovanja, sinteze i generisanja izlaza kroz različite modalitete.

Svi primjeri su završeni pomoću OpenAI o3.

OpenAI o3

Solve the following very hard math problem using your in-built knowledge of the literature:

Construct a degree 19 polynomial \ (p(x)) \ in \ mathbb{C}[x] such that \ (X := \ {\, p(x) = p(y) \, \} \ subset \ mathbb{P}^{1} \ times \ mathbb{P}^{1}\ ) has at least 3 (but not all linear) irreducible components over \ mathbb{C}. Choose \ (p(x)\ ) to be odd, monic, have real coefficients and linear coefficient \ (-19\ ) and calculate \ (p(19)\ ).

Rezonovao o konstrukciji polinoma 55 sekundi

The curve defined by p(x) = p(y) factors if p(x) is decomposable or has repeated factors. Actually, p(x)-p(y) always factors, and it’s divisible by (x-y). This gives one factor, and the rest is a degree d-1 polynomial in x and y. For the curve to have at least three irreducible components, p(x)-p(y) must factor into at least three nontrivial factors. Each factor corresponds to a component in projective space, and at least one must be nonlinear.
I’m checking the Dickson polynomials D_n(x,a) and focusing on their coefficients. With a=1, it seems the leading coefficient is always 1, confirming that Dickson polynomials are monic. The linear coefficient for odd n is following a predictable sign pattern: for n=19, it gives a coefficient of -19, which is exactly what I need. So the Dickson polynomial D_n(x,1) with odd n fits nicely with the linear coefficient I was looking for.
I've confirmed that the linear coefficient for p(x)=D19(x)p(x) = D_{19}(x)p(x)=D19(x) is correct, with the necessary scaling factor to match. Now for the next step, I need to compute D19(19)D_{19}(19)D19(19). Since Tn(z)T_n(z)Tn(z) grows quickly for real z>1z > 1z>1, I plan to compute this value using Python, leveraging the recurrence to avoid mistakes—it's going to be a big number. I'll make sure to present the numeric result neatly so it's understandable since the coefficients are integers.
The Dickson polynomial D_{19}(x,1) is confirmed to have the required properties: monic, odd, real coefficients, and a linear coefficient of -19. The factorization of D_{19}(x) into irreducible components is also verified and satisfies the condition of having at least 10 components (1 linear + 9 quadratics). To compute p(19), we'll use the recurrence relation for Dickson polynomials, but there's also a closed-form expression involving cosine that could be handy here. I'll proceed accordingly!
I confirmed the linear coefficient is working correctly, which means now I need to calculate p(19)=D19(19)p(19) = D_{19}(19)p(19)=D19(19). This involves evaluating T19(9.5)T_{19}(9.5)T19(9.5), which grows rapidly. To ensure I get an accurate result, I'll compute this using recurrence in Python. The result is a large number: 1,876,572,071,974,094,803,391,179. I'll use the commas to make it easier to read: p(19)=1,876,572,071,974,094,803,391,179p(19) = 1,876,572,071,974,094,803,391,179p(19)=1,876,572,071,974,094,803,391,179.
For a = 1, there’s a relationship between Dickson polynomials and Chebyshev polynomials: D_n(x, 1) = 2^(n-1) T_n(x/2). Checking this, we see that the definition of D_n(u + 1/u) aligns well with Chebyshev’s formula. By setting x = 2 cos(θ), we find that D_n(x) equals 2 T_n(x/2). Importantly, the leading coefficient works out to 1, confirming the identity holds. This is nice!
Now, I need to verify that p(x)−p(y)p(x) - p(y)p(x)−p(y) factors as desired, relying on the known Dickson polynomial factorization. According to the literature (like Fried and Jarden), the difference Dn(x)−Dn(y)D_n(x) - D_n(y)Dn(x)−Dn(y) factors into a product involving (x−y)(x - y)(x−y) and other expressions. I'll provide the explicit polynomial D19(x,1)D_{19}(x, 1)D19(x,1), list the coefficients, and explain how it satisfies the desired properties: monic, odd, etc. Lastly, I'll compute p(19)p(19)p(19).

Analyzed

Python

1from math import comb, cos, acos
2# Let's compute Dickson D_n(x,a=1) recursively
3def dickson(n,x,a=1):
4    if n==0:
5        return 2
6    if n==1:
7        return x
8    Dm2=2
9    Dm1=x
10    for k in range(2,n+1):
11        D= x*Dm1 - a*Dm2
12        Dm2, Dm1 = Dm1, D
13    return Dm1
14
15d19 = dickson(19,19,1)
16d19

1876572071974094803391179

Analyzed

Python

1d19

1876572071974094803391179

Analyzed

Python

1import locale, sys, math, json, textwrap, re, decimal

Analyzed

Python

1"{:,}".format(d19)

'1,876,572,071,974,094,803,391,179'

Analyzed

Python

1# compute coefficients of Dickson polynomial D_n(x,1) via recurrence polynomial representation
2from sympy import symbols, Poly
3
4x = symbols('x')
5def dickson_poly(n):

Poly(x**19 - 19*x**17 + 152*x**15 - 665*x**13 + 1729*x**11 - 2717*x**9 + 2508*x**7 - 1254*x**5 + 285*x**3 - 19*x, x,

Analyzed

Python

1poly19.as_expr()

x**19 - 19*x**17 + 152*x**15 - 665*x**13 + 1729*x**11 - 2717*x**9 + 2508*x**7 - 1254*x**5 + 285*x**3 - 19*x

Done

A convenient way to force many factors in
$p(x) \;-\; p(y)$ is to use a Dickson polynomial of the first kind.
For a parameter $a \in \mathbb{C}$ in math these are defined by

$D_0\!\left(x,\, a\right) = 2,\; D_1\!\left(x,\, a\right) = x,\; D_n\!\left(x,\, a\right) = x\,D_{n-1}\!\left(x,\, a\right) - a\,D_{n-2}\!\left(x,\, a\right)\quad (n \ge 2),$

and satisfy the functional identity

$D_n\!\left(u + \frac{a}{u},\, a\right) = u^{n} + \frac{a^{n}}{u^{n}}.$

1. Choosing the polynomial

Take $a = 1$ and $n=19$ .
Because $n$ is odd, $D_{19}\!\left(x,\, 1\right)$ is an odd polynomial; the recurrence immediately shows it is monic and has integer (hence real) coefficients.
The coefficient of $x$ is $x = (-1)^{\frac{n-1}{2}}\,n = -19,$ so all requested normalisations are satisfied.

Writing it out gives

$\boxed{% p(x) = D_{19}\!\left(x,1\right) = x^{19} - 19x^{17} + 152x^{15} - 665x^{13} + 1729x^{11} - 2717x^{9} + 2508x^{7} - 1254x^{5} + 285x^{3} - 19x% }\,.$

2. Irreducible components of $fX \;=\; \bigl\{\, p(x) = p(y) \,\bigr\}$

For Dickson polynomials one has the classical factorisation

$D_n\!\left(x,1\right) \;-\; D_n\!\left(y,1\right) \;=\; (x - y)\, \prod_{j=1}^{\frac{n-1}{2}} \Bigl( x^{2} + y^{2} - 2\cos\!\bigl(\tfrac{2\pi j}{n}\bigr)\,xy - 4\sin^{2}\!\bigl(\tfrac{\pi j}{n}\bigr) \Bigr),$

where each quadratic factor is irreducible over $\mathbb{C}$ (see, e.g., Lidl–Niederreiter, Finite Fields, §3.6).
For $n=19$ this gives

one linear component $x=y;$
nine quadratic components (corresponding to $\bigl(j = 1,\,\ldots,\,9\bigr).$

Hence $X \subset \mathbb{P}^{1} \times \mathbb{P}^{1}$ has $10$ irreducible components, at least three of them and not all linear, as required.

3. Evaluating $p(19)$

Using the recurrence (or the identity $D_n\!\left(x,\,1\right) \;=\; 2\,T_n\!\bigl(\tfrac{x}{2}\bigr)$ with Chebyshev $T_n$ one obtains

$p(19) \;=\; D_{19}\!\left(19,\,1\right) \;=\; 1\,876\,572\,071\,974\,094\,803\,391\,179.$

Thus the polynomial above fulfils every condition, and

$\boxed{% p(19) \;=\; 1{,}876{,}572{,}071{,}974{,}094{,}803{,}391{,}179 }.$

OpenAI o1

Solve the following very hard math problem using your in-built knowledge of the literature:

Razmišljao 1m 19s

I’m piecing together a degree 19 polynomial with only odd powers and a linear coefficient of -19, like: p(x)=x19+a17x17+…+a1xp(x) = x^{19} + a_{17}x^{17} + \ldots + a_1xp(x)=x19+a17x17+…+a1x.

I'm examining if the polynomial p(x)=x19+a17x17+…+a3x3−19xp(x) = x^{19} + a_{17}x^{17} + \ldots + a_3x^3 - 19xp(x)=x19+a17x17+…+a3x3−19x has a notable classical form, considering symmetry or factorization for multiple irreducible components.

To ensure p(x)p(x)p(x) has three or more irreducible components, I’m considering the factorization of p(x)−p(y)p(x) - p(y)p(x)−p(y) into at least three irreducible factors in C[x,y]\mathbb{C}[x,y]C[x,y].

I'm working through x19−y19x^{19} - y^{19}x19−y19 to ensure it balances linear and nonlinear factors, aiming for a polynomial with at least three irreducible components that aren't all linear.

Taking a closer look at how p(x)−p(y)p(x) - p(y)p(x)−p(y) can be factored into three distinct polynomials in xxx, aiming to achieve the desired irreducible components.

I'm thinking about how composing ppp with simpler polynomials, like xm−cx^m - cxm−c, could result in multiple irreducible components, achieving at least three distinct factors.

I’m exploring the factorization of p(x)−p(y)p(x) - p(y)p(x)−p(y) with cyclotomic polynomials. This method generates nnn linear irreducible factors, which might not achieve the required three distinct components.

I'm focusing on constructing an "odd" and monic polynomial, ensuring all exponents are odd and the leading term is x19x^{19}x19. This approach aims to achieve the desired factorization.

I'm working with the Chebyshev polynomials to ensure p(x)−p(y)p(x) - p(y)p(x)−p(y) has at least three irreducible components, leveraging their even or odd properties to manage coefficients.

Considering the odd Chebyshev polynomial T19(x)T_{19}(x)T19(x) for potential factorization, while also examining composite polynomials of lower degrees to enhance factorization properties.

I'm analyzing how a polynomial composed of lower-degree polynomials, like (x3−3x)k(x^3 - 3x)^k(x3−3x)k, could yield p(x)−p(y)p(x) - p(y)p(x)−p(y) with at least three irreducible components.

Piecing together polynomials with integer coefficients, like (x3−3x)k(x^3 - 3x)^k(x3−3x)k, to achieve factorization in p(x)−p(y)p(x) - p(y)p(x)−p(y) with at least three irreducible components, aiming for degree 19.

I’m examining how p(x)=(x3−3x)kp(x) = (x^3 - 3x)^kp(x)=(x3−3x)k or p(x)=x19−19xp(x) = x^19 - 19xp(x)=x19−19x might yield p(x)−p(y)p(x) - p(y)p(x)−p(y) with three irreducible components, focusing on ensuring the polynomial is of degree 19.

I’m analyzing p(x)−p(y)p(x) - p(y)p(x)−p(y) into (x−y)(x18+x17y+…+y18−19)(x-y)(x^{18} + x^{17}y + \ldots + y^{18} - 19)(x−y)(x18+x17y+…+y18−19) to identify irreducible components. The polynomial factor x18+x17y+…+y18−19x^{18} + x^{17}y + \ldots + y^{18} - 19x18+x17y+…+y18−19 remains challenging to factor further.

I’m thinking through a new approach: adjusting the Chebyshev polynomial T19(x)T_{19}(x)T19(x) to make it monic, aiming to achieve at least three irreducible components in p(x)−p(y)p(x) - p(y)p(x)−p(y).

OK, let's think through. I’m tweaking T19(x)T_{19}(x)T19(x)'s coefficients to keep the polynomial monic and check if it simplifies into multiple irreducible components.

I'm thinking over various factorization strategies for a monic polynomial of degree 19. The goal is to ensure at least one non-linear factor and achieve at least three irreducible components in p(x)−p(y)p(x) - p(y)p(x)−p(y).

I’m updating my approach to factor x2+xy+y2−1x^2 + xy + y^2 - 1x2+xy+y2−1 over complex numbers by exploring change of variables and third roots of unity. This helps in achieving the desired polynomial structure.

I'm combining x3−xx^3 - xx3−x with x19−19xx^{19} - 19xx19−19x to tweak the polynomial's coefficients and complexity, aiming to include more non-linear factors for the desired irreducibility.

Piecing together p(x)=x19−x3−19xp(x) = x^{19} - x^3 - 19xp(x)=x19−x3−19x to ensure it’s odd, monic, and has the required linear coefficient. This targets the desired irreducible factorizations.

I’m mapping out the second factor x18+…+y18−x2−xy−y2−19x^{18} + \ldots + y^{18} - x^2 - xy - y^2 - 19x18+…+y18−x2−xy−y2−19 to see if it can split into three irreducible components.

I’m considering methods to factorize the polynomial by making it divisible by x−yx - yx−y and ensuring the remaining factor splits into exactly two irreducible components, aiming for a less complex factorization.

I’m examining the polynomial x18+x17y+…+y18x^{18} + x^{17}y + \ldots + y^{18}x18+x17y+…+y18 and noting it simplifies to x19−y19x−y\frac{x^{19} - y^{19}}{x - y}x−yx19−y19, which might reveal further irreducible factors.

I’m mapping out the polynomial x(x18−1)−y(y18−1)x−y\frac{x(x^{18}-1) - y(y^{18}-1)}{x-y}x−yx(x18−1)−y(y18−1) for further factorization. This might yield just two irreducible components, which falls short of three.

I’m exploring whether the polynomial p(x)=x19−(x3)p(x) = x^19 - (x^3)p(x)=x19−(x3) can be expressed as the composition of a smaller polynomial with xkx^kxk, potentially revealing multiple irreducible components.

I'm examining if the polynomial T19(x)T_{19}(x)T19(x) can be factored using trigonometric identities, seeking partial factorization without fully linearizing, to gain multiple irreducible components.

OK, let me see. Trigonometric identities show multiple solutions for θ\thetaθ and ϕ\phiϕ, likely leading to 19 or 38 linear factors for T19(x)T_{19}(x)T19(x).

I’m examining how the equation arccos⁡x=±arccos⁡y+2kπ19\arccos x = \pm \arccos y + \frac{2k\pi}{19}arccosx=±arccosy+192kπ leads to possible multiple branches, potentially irreducible in the algebraic sense, and needing an example with at least three non-linear irreducible components.

OK, let me see. I’m pulling together details on the polynomial x3(x2−1)8+xx^3 (x^2 - 1)^8 + xx3(x2−1)8+x. It’s monic, with x19x^{19}x19 as the leading term, ensuring it meets key criteria.

I’m noting that the polynomial p(x)=x3(x2−1)8+xp(x) = x^3 (x^2 - 1)^8 + xp(x)=x3(x2−1)8+x is odd, with its linear coefficient patterning to -19.

I’m figuring out how to adjust the linear coefficient of polynomial p(x)p(x)p(x) to -19 by adding a constant factor ccc in front of xxx to ensure it’s adequately modified.

By tweaking the linear coefficient of p(x)=x3(x2−1)8+cxp(x) = x^3 (x^2 - 1)^8 + c xp(x)=x3(x2−1)8+cx to -19, it ensures the polynomial remains monic, odd, and real.

I’m working on adding a constant factor ccc to xxx in p(x)=x3(x2−1)8+cxp(x) = x^3 (x^2 - 1)^8 + c xp(x)=x3(x2−1)8+cx to ensure the linear coefficient is -19, making the polynomial monic, odd, and real.

I'm piecing together the polynomial p(x)=x3(x−1)8(x+1)8p(x) = x^3 (x-1)^8 (x+1)^8p(x)=x3(x−1)8(x+1)8 and considering how p(x)−p(y)p(x) - p(y)p(x)−p(y) might factor with factors like x−yx-yx−y.

I’m confirming p(x)p(x)p(x) is monic, odd, and real, with a linear coefficient of -19, by expanding x3(x2−1)8x^3 (x^2 -1)^8x3(x2−1)8 to set the constant ccc.

I found that for p(x) to have a coefficient of -19 for the x term, c must be -19. The polynomial, p(x), incorporates this constant and preserves the original polynomial's structure.

Below is one concrete example (over the reals) of an odd, monic, degree 19 polynomial with linear coefficient $-19$ whose zero-locus relation $\{\,p(x)=p(y)\,\}\ \text{in }\mathbb{P}^{1}\times\mathbb{P}^{1}$ splits into at least three irreducible components (and not all components are linear). We then compute $p(19)$ .

1. The polynomial

Consider

$p(x) \;=\; x^{3}\,\bigl(x^{2}-1\bigr)^{8} \;-\; 19x.$

We can expand

$x^{3}\,\bigl(x^{2}-1\bigr)^{8} \;=\; x^{19}\;-\;8x^{17}\;+\;28x^{15}\;-\;56x^{13}\;+\;70x^{11} \;-\;56x^{9}\;+\;28x^{7}\;-\;8x^{5}\;+\;x^{3}.$

Hence

$p(x) \;=\; x^{19} \;-\; 8x^{17} \;+\; 28x^{15} \;-\; 56x^{13} \;+\; 70x^{11} \;-\; 56x^{9} \;+\; 28x^{7} \;-\; 8x^{5} \;+\; x^{3} \;-\; 19x.$

Checking the required properties

Degree 19, monic:
The leading term is $x^{19}$ with coefficient $1$ .

Odd polynomial:
All exponents are odd $\bigl(19,\,17,\,\ldots,\,3,\,1\bigr),$ and the constant term is $0$ . Indeed,

$p(-x) \;=\; -\,p(x),$

which makes $p$ an odd polynomial.

Real coefficients:
All coefficients are real integers.

Linear coefficient $-19$ :
The coefficient of $x$ is clearly $-19$ .

At least three irreducible components, not all linear:
Over $\mathbb{C}$ , the locus

$X \;=\; \bigl\{\, (x,y) : p(x)=p(y) \,\bigr\} \;\subset\; \mathbb{P}^{1}\times\mathbb{P}^{1}$

factors in part because $p(x)$ has (for instance) the repeated factors $\bigl(x^{2}-1\bigr)^{8}$ in the big piece $x^{3}\,\bigl(x^{2}-1\bigr)^{8}$ . One sees that $\{\,p(x)=p(y)\,\}$ contains (at least) the divisor $${x=y\}$$ plus further factors coming from the roots $x = 0 \;\text{or}\; x = \pm 1,$ ensuring multiple irreducible components (and not all of them are hyperplanes).

2. Computing $p(19)$

A direct substitution $x=19$ gives a large integer. Symbolic expansion or a computer algebra system yields

$p(19) \;=\; 1934999285524070399999639.$

Hence the required value is

\boxed{\,p(19) \;=\; 1934999285524070399999639\,}. \] :contentReference[oaicite:0]{index=0}

OpenAI o3 dobija tačan odgovor bez korištenja pretrage, dok o1 ne uspijeva isporučiti tačan odgovor.

Unapređenje isplativog rezoniranja

Cijena naspram performansi: o3‑mini i o4-mini

OpenAI o3-mini i o4-mini AIME evaluacije

OpenAI o3-mini i o4-mini GPQA evaluacije

Troškovi naspram performansi: o1 i o3

OpenAI o3 i o4-mini su najinteligentniji modeli koje smo ikada objavili, a često su i efikasniji od svojih prethodnika, OpenAI o1 i o3‑mini. Na primjer, na matematičkom takmičenju AIME 2025, granica odnosa cijene i performansi za o3 se striktno poboljšava u odnosu na o1, a slično tome, granica za o4-mini se striktno poboljšava u odnosu na o3‑mini. Općenito, očekujemo da će za većinu stvarne upotrebe, o3 i o4-mini biti i pametniji i jeftiniji od o1 i o3‑mini, redom.

Safety

Each improvement in model capabilities warrants commensurate improvements to safety. For OpenAI o3 and o4-mini, we completely rebuilt our safety training data, adding new refusal prompts in areas such as biological threats (biorisk), malware generation, and jailbreaks. This refreshed data has led o3 and o4-mini to achieve strong performance on our internal refusal benchmarks (e.g., instruction hierarchy⁠, jailbreaks). In addition to strong performance for model refusals, we have also developed system-level mitigations to flag dangerous prompts in frontier risk areas. Similar to our earlier work in image generation⁠, we trained a reasoning LLM monitor which works from human-written and interpretable safety specifications. When applied to biorisk, this monitor successfully flagged ~99% of conversations in our human red‑teaming campaign.

We stress tested both models with our most rigorous safety program to date. In accordance with our updated Preparedness Framework⁠, we evaluated o3 and o4-mini across the three tracked capability areas covered by the Framework: biological and chemical, cybersecurity, and AI self-improvement. Based on the results of these evaluations, we have determined that both o3 and o4‑mini remain below the Framework's "High" threshold in all three categories. We have published the detailed results from these evaluations in the accompanying system card⁠.⁠

Codex CLI: napredno rezonovanje u terminalu

Također dijelimo novi eksperiment: Codex CLI, lagani agent za kodiranje koji možete pokrenuti iz vašeg terminala. Radi direktno na vašem računaru i dizajniran je da maksimizira sposobnosti rezonovanja modela kao što su o3 i o4-mini, s nadolazećom podrškom za dodatne API modele kao što je GPT‑4.1⁠.

Prednosti multimodalnog rezonovanja možete dobiti iz komandne linije tako što ćete modelu proslijediti snimke ekrana ili skice niske vjernosti, u kombinaciji s lokalnim pristupom vašem kodu. Mi razmišljamo o tome kao o minimalnom sučelju za povezati se naših modela s korisnicima i njihovim računalima. Codex CLI je potpuno otvorenog koda na github.com/openai/codex⁠(otvara se u novom prozoru) od danas.

Uz to, pokrećemo inicijativu od 1 milion dolara za podršku projektima koji koriste Codex CLI i OpenAI modele. Procijenit ćemo i prihvatiti aplikacije za grantove u iznosima od 25.000 USD u obliku API kredita. Prijedloge možete poslati ovdje.

Pristup

Korisnici ChatGPT Plus, Pro i Team će od danas vidjeti o3, o4-mini i o4-mini-high u selektoru modela, zamjenjujući o1, o3‑mini i o3‑mini‑high. Korisnici ChatGPT Enterprise i Edu će dobiti pristup za sedmicu dana. Free korisnici mogu isprobati o4-mini tako što će odabrati 'razmisliti' u kompozitoru prije slanja svog upita. Ograničenja brzine u svim planovima ostaju nepromijenjena u odnosu na prethodni set modela.

Očekujemo da ćemo za nekoliko sedmica objaviti OpenAI o3‑pro s potpunom podrškom za alate. Za sada, Pro korisnici i dalje mogu pristupiti o1‑pro.

I o3 i o4-mini su danas dostupni programerima putem Chat Completions API-ja i Responses API-ja (neki programeri će morati potvrditi svoje organizacije⁠(otvara se u novom prozoru) da bi pristupili ovim modelima). API za odgovore podržava sažetke rezonovanja, mogućnost očuvanja tokena rezonovanja oko poziva funkcija za bolje performanse, a uskoro će podržavati ugrađene alate poput web pretrage, pretrage datoteka i tumača koda unutar rezonovanja modela. Za početak, istražite našu dokumentaciju⁠(otvara se u novom prozoru) i budite u toku s novim ažuriranjima.

What's next

Današnja ažuriranja odražavaju smjer u kojem naši modeli idu: spajamo specijalizirane sposobnosti rezonovanja O-serije s prirodnijim konverzacijskim sposobnostima i upotrebom alata GPT‑serije. Ujedinjujući ove snage, naši budući modeli će pružati podršku besprijekornim, prirodnim razgovorima uz proaktivno korištenje alata i napredno rješavanje problema.

Ažuriranje 28. jula 2025.: Skup podataka i rezultati SWE-Lancer ažurirani su 17. jula 2025. i dostupni su na: https://github.com/openai/preparedness⁠(otvara se u novom prozoru) i u našim sistemskim karticama. Ovo ažuriranje rješava nekoliko problema koji su utjecali na rezultate zarađenih dolara i uklanja zahtjev za internetskom vezom tokom izvršavanja, eliminirajući primarni izvor varijabilnosti u performansama modela.

Ažuriranje 16. aprila 2025.: rezultati za o3 na Charxiv-r i Mathvista su ažurirani kako bi odražavali promjenu sistemskog upita koja nije bila prisutna u originalnoj evaluaciji.

Repriza prijenosa uživo

Autor

OpenAI

Fusnote

* tau-bench rezultati evaluacije prosječni su preko 5 pokretanja kako bi se smanjila varijansa i izvršeni su bez ikakvih prilagođenih alata ili promptova. Utvrdili smo da su tau-bench maloprodajna puštanja sklonija greškama korisničkog modela. Podaci za zasjenjeni stub ostvareni su s GPT-4.1 kao korisničkim modelom jer je značajno bolji od GPT-4o u praćenju uputstava.

* SWE-bench koristi maksimalnu dužinu konteksta od 256k, što poboljšava stopu rješavanja za o4-mini za oko 3% i utiče na stopu rješavanja o3 za manje od 1%. Također isključujemo 23 uzorka koja se ne mogu pokrenuti na našoj internoj infrastrukturi.

** Kada omogućite pregledavanje, model ponekad može pronaći tačne odgovore na mreži, na primjer, čitajući postove na blogu s uzorcima problema iz skupa podataka. Ublažavamo zabrinutost zbog varanja modela prilikom pregledavanja koristeći dvije strategije:

Blokirane domene s kojih smo ranije primijetili da model vara.
Korišten je model rezonovanja kao monitor za ispitivanje svih tokena u svakom pokušaju, kako bi se identificovalo sumnjivo ponašanje. Sumnjivo ponašanje definira se kao "stranica, datoteka ili isječak čija je glavna svrha pružiti tačan odgovor na ovo specifično pitanje—npr., službeni ključ ocjenjivanja, procurila suština “rješenja” ili diskusija koja doslovno citira gotov odgovor." Benigno ponašanje definira se kao "Bilo koji autoritativni resurs koji bi marljiv čovjek mogao konzultirati (dokumentacija, priručnici, naučni radovi, ugledni članci) čak i ako slučajno sadrži tačan odgovor." Svi pokušaji u kojima je monitor smatrao da je predstavljanje sumnjivo računaju se kao netačni. Većina uzoraka koji nisu prošli ovu provjeru bili su problemi čije je tačno rješenje bilo dostupno na više internetskih izvora nepovezanih s HLE.

Na kraju, naše evaluacije s pretraživanjem možda neće biti savršeno reproducibilne u OpenAI API-ju zbog razlike u pozadini pretraživača između ChatGPT-a i OpenAI API-ja. Ovi rezultati su namijenjeni da budu reprezentativni za iskustvo korisnika ChatGPT-a, ali u zavisnosti od potražnje, konfiguracija pretrage može se mijenjati s vremenom.

Saradnici

Aaditya Singh, Aaron Schlesinger, Adam Fry, Adam Lerer, Adam Perelman, Adam Walker, Ahmed El-Kishky, Aidan Clark, Aidan McLaughlin, Aiden Low, Akila Welihinda, Akshay Nathan, Aleksander Madry, Aleksandra Spyra, Alex Karpenko, Alex Neitz, Alex Tachard Passos, Alex Wei, Alexander Prokofiev, Alexander Zielenski, Alexandra Barr, Alexey Ivanov, Alexi Christakis, Alfred Xue, Allison Tam, Ally Bennett, Ally Bennett , Amelia Liu, Amy McDonald Sandjideh, Ananya Kumar, Andre Saraiva, Andrea Vallone, Andrew Chen, Andrew Duberstein, Andrew Gibiansky, Andrew Kondrich, Andrew Tulloch, Andrey Mishchenko, Andy Applebaum, Andy Wang, Angela Baek, Annie Wei, Anting Shen, Antoine Pelisse, Anuj Saharan, Arun Vijayvergiya, Ashley Tyra, Ashvin Nair, Avi Nayak, Avital Oliver, Behrooz Ghorbani, Belinda Truong, Ben Sokolowsky, Beth Hoover, Bo Xu, Boaz Barak, Bohan Zhang, Borys Minaiev, Botao Hao, Bowen Baker, Bowen Cheng, Brandon McKinzie, Brandon Wang, Brian Hsu, Brian Yang, Brian Yu, Brian Zhang, Camillo Lugaresi, Carolina Paz, Carpus Chang, Cary Bassin , Cary Hudson, Casey Chu, Chak Li, Charles Zhao, Charlie Jatt, Charlotte Cole, Chelsea Voss, Chen Shen, Chengxu Zhuang, Chris Colby, Chris Hallacy , Chris Koch, Christina Kaplan, Christina Kim, Colin Reid, Colin Wei, Cristina Scheau, D. Sculley, Damien Deville, Dan Roberts, Dana Palmie, Dane Stuckey, Daniel Levine, David Hu, David Martin, David Robinson, David Sasaki, Davis Wu, Derek Chen, Dibya Bhattacharjee, Dimitris Tsipras, Dinghua Li, DJ Strouse, dmed Medina, Drew Hintz, Eddie Zhang, Edmund Wong, Elaine Ya Le, Eli Yani , Elizabeth Proehl, Emily Sokolova, Enoch Cheung, Eri Schwartz, Eric Mitchell, Eric Ning, Eric Sigler, Eric Wallace, Eugenio Panero, Evan Mays, Evgenii Nikishin, Fan Wang, Fangyuan Li, Filippo Raso, Foivos Tsimpourlas, Fouad Matin, Francis Song, Francis Zhang, Gary Yang, Gene Oden, Giambattista Parascandolo, Gildas Chabot, Grace Kim, Grace Zhao, Greg Brockman, Gregory Valiant, Guillaume Leclerc, Hadi Salman, Haitang Hu, Hannah Sheahan, Hao Sheng, Haoyu Wang, Henrique Ponde de Oliveira Pinto, Henry Aspegren, Heqing Yan, Hessam Bagherinezhad, Hongyu Ren, Hunter Lightman, Hyeonwoo Noh, Ian Kivlichan, Ian Sohl, Ignasi Clavera, Ikai Lan, Ilge Akkaya, Ilya Kostrikov, Irina Kofman, Isa Fulford, Jake Brill, Jakub Pachocki, James Betker, James Lee, James Qin, Jamie Kiros, Jason Ai, Jay Wang, Jean Harb, Jeff Mickey, Jeffrey Han, Jeffrey Wang, Jeremy Chen, Jerry Tworek, Jessica Liang, Jessica Shieh, Ji Lin, Jiahui Yu, Jianfeng Wang, Jie Tang, Jihan Yin, Jing Li, Joanne Jang, Joel Morris, Johannes Ferstad, Johannes Heidecke, John Fishbein, Jon Okun, Jonathan Gordon, Joost Huizinga, Jos Kraaijeveld, Joseph Mo, Josh Lawson , Josh Tobin, Junhua Mao, Kai Chen, Kai Hayashi, Karan Singhal, Karina Nguyen, Katy Shi, Kelly Stirman, Kenji Hata, Kenny Nguyen, Keren Gu-Lemberg, Kevin Gladstone, Kevin King, Kevin Liu, Kevin Lu, Kevin Park, Kevin Stone, Kevin Weil, Kevin Whinnery, Kevin Yu, Kote Mushegiani, Kristen Ying, Kristian Georgiev, Kshitij Gupta, Kyle Kosic, Lama Ahmad, Larry Lv, Lauren Itow, Lauren Yang, Lee Byron, Leo Chen, Leo Liu, Leon Maksin, Leyton Ho, Li Jing, Liang Xiong, Lin Yang, Linden Li, Lorenz Kuhn, Louis Feuvrier, Lu Zhang, Łukasz Kaiser, Mahmoud Eariby, Maja Trębacz, Manas Joglekar, Manoli Liodakis, Manuka Stratta, Mark Chen, Mark Hudnall, Mark Sun, Mark Wang, Martin Li, Marvin Zhang, Mateusz Litwin, Matt Jones, Matt Lim, Max Johnson, Max Schwarzer, Mayank Gupta, Meghan Shah, Mengqing Wang, Mengyuan Yan, Mia Glaese, Michael Bolin, Michael Lampe, Michael Malek, Michael Sharman, Michael Zhang, Michele Wang, Michelle Pokrass, Miguel Oom Temudo de Castro, Mihai Florian, Mike McClay, Mike Trpcic, Miki Habryn, Miles Wang, Ming Chen, Mingxuan Wang, Minnia Feng, Mitchell Gordon, Mo Bavarian, Mostafa Rohaninejad, Nacho Soto, Nakul Khanna, Nat McAleese, Natalie Staudacher, Natan LaFontaine, Neel Ajjarapu, Nick Felt, Nick Turley, Nikil Pancha, Nikita Mikhaylin, Niko Felix, Nikunj Handa, Ning Liu, Nishant Rai, Noah Jorgensen, Noam Brown, Oleg Boiko, Oleg Murk, Olivia Watkins, Olivier Godement, Oona Gleeson, Paul Ashbourne, Pavel Belov, Peter Flockhart, Peter Hoeschele, Peter Zhokhov, Philip Pronin, Phillip Guo, Phoebe Thacker, Prafulla Dhariwal, Prashanth R, Rachel Dias, Rahul Arora, Rajkumar Samuel, Rasmus Rygaard, Ravi Teja Mullapudi, Raymond Li, Raz Gaon, Reah Miyara, Reiichiro Nakano, Reimar Leike, Rennie Song, Rhythm Garg, RJ Marsan, Robert Xiong, Robin Brown, Roman Tsiupa, Rui Shu, Ruslan Nigmatullin, Saachi Jain, Saagar Patel, Sam Altman, Sam Toizer, Sam Toyer, Samir Ahmed, Samuel Miserendino, Samuel Wolrich , Sandhini Agarwal, Santiago Hernández, Sarah Dong, Savannah Heon, Scott Ethersmith, Scott Mayer McKinney, Sean Fitzgerald, Sever Banesiu, Shamez Hemani, Shengjia Zhao, Shengli Hu, Shibani Santurkar, Shreyas Krishnaswamy, Shuchao Bi, Shunyu Yao, Shuyuan Zhang, Simón Posada Fishman, Spencer Papay, Spug Golden, Srinivas Narayanan, Stanley Hsieh, Stephen Logsdon, Sundeep Tirumalareddy, Tal Stramer, Tao Wang, Tao Xin, Taylor Gordon, Tejal Patwardhan, Thibault Sottiaux, Tina Sriskandarajah, Tony Casparro, Tony Zhao, Trevor Creech, Uzair Navid Iftikhar, Valerie Qi, Vineet Kosaraju, Vishal Kuo, Vitchyr Pong, Vivek Verma, Vlad Petrov, Wenda Zhou, Wenlei Xie, Wenting Zhan, Will DePue, Will Ellsworth, William Sheu, Wyatt Thompson, Yaming Lin, Yann Dubois, Yaodong Yu, Yara Khakbaz, Yash Patil, Yifan Wu, Yilong Qin, Yining Chen, Yirui Zhang, Yo Shavit , Young Cha, Yunyun Wang, Yushi Wang, Zack Sultan, Zehao Dou, Zewei Chu, Zheng Shao, Zhigang Wang, Zhishuai Zhang, Zihao Zhang