OpenAI’s12 Days of OpenAIcampaign wrapped up today with two major announcement : they unveil o3 and o3 Mini models . Just days after the full expiration of the o1 mannikin , OpenAI has outdo itself by announcing an even well - performing abstract thought model . Here ’s everything you necessitate to know .
board of Contents
1. o3 Reasoning Model
OpenAI ’s o series AI modelling are abstract thought models , have in mind they take fourth dimension to think stair - by - footstep to get in at a conclusion . This advance enable better truth and allows these models to solve complex problem , peculiarly in scheduling , math , and science .
The full version of theo1 modelling was releasedon Day 1 of the “ 12 Days of OpenAI ” campaign . However , as challenger intensifies with models likeGemini 2.0 Flash Thinking , OpenAI has lift the bar with the o3 model , which demonstrate meaning advance over the o1 fashion model in both benchmarks and accuracy .
For example , in programming job , the model ’s truth improved from 48.9 % to 71.7 % .
The o3 good example also designate renowned onward motion in solving math problems and Ph.D. - story science questions .
to boot , it achieved a significant improvement in theARC Prize benchmark(Abstraction and Reasoning Corpus for Artificial General Intelligence ) . Basically , it try out an AI ’s ability to puzzle out unfamiliar puzzles using system of logic and normal recognition . Instead of rely on previous training or data point , the benchmark focuses on how well the AI can conceive and vulgarise to solve unexampled problems , similar to human reasoning .
Today OpenAI announced o3 , its next - gen reasoning model . We ’ve worked with OpenAI to test it on ARC - AGI , and we believe it represents a significant find in getting AI to adapt to fresh tasks . It scores 75.7 % on the semi - individual eval in dispirited - compute mood ( for $ 20 per task…pic.twitter.com/ESQ9CNVCEA
2. o3 Mini: Cost-Efficient Model
Along with the o3 mannequin , OpenAI also declare o3 Mini , a model designed for speed and efficiency .
o3 Mini focus on low , medium , and high - abstract thought task , offer a counterbalance of capability and cost - effectiveness . In programming benchmarks , while o3 Mini ( Low ) achieve the same Elo rating as the o1 Mini , the o3 Mini ( High ) outperform even the full o1 fashion model .
A similar style was observed in maths bench mark . However , it ’s worth noting that while o3 Mini ( Low ) having the same accuracy of the o1 Mini , it achieves this with greater efficiency , subjugate the latent period of its output .
During the hot demo , o3 Mini prove its versatility by writing scripts that evaluated its own performance .
Also Read :
Availability
Both o3 and o3 Mini models are not yet usable to the world . OpenAI is prioritizing safety , granting other access exclusively to pick out research worker for outside safety testing . Applications for international prophylactic examination opened today , and you canapply from here .
While a specific sacking date was not provide , o3 Mini is carry to launch presently after the discharge of o3 .
The End of 12 Days of OpenAI
OpenAI ’s campaign resolve today , and among all the declaration , Sora , its text - to - video recording generation example , stands out as the most exciting . Updates toCanvasandSearchare also extremely practical and beneficial for a panoptic compass of exploiter .
While the updates to the reasoning models , such as o3 and o3 Mini , interpret a significant leaping in AI capacity , they may be less relevant for users who are not into solving Ph.D. - storey problem .