OpenAI says its reasoning system solved all 12 problems at the 2025 ICPC World Finals; GPT-5 solved 11 and an experimental model solved #12 after GPT-5 couldn't
An OpenAI system has solved every problem at the world's most prestigious collegiate programming championship …
The DecoderMaximilian Schreiner
Context & Ripple Effects
This result extends OpenAI's recent push around extended reasoning: GPT-5 Pro was introduced as an extended-reasoning tier, while developers had reported GPT-5 was stronger at technical planning even as other models could still produce better code in some cases.
The ICPC claim is a sharper competitive-programming test of that trajectory. It also contrasts with earlier research finding GPT-3.5-based systems struggled more with newer coding problems, making progress on hard, unfamiliar problem-solving the salient arc.
First-order effects
OpenAI gains a high-visibility demonstration for its reasoning-and-coding systems, though the reported sweep belongs to a system combining GPT-5 with an experimental model rather than GPT-5 alone.
The result identifies a concrete boundary for GPT-5: it solved 11 problems, while a separate experimental model was needed for the remaining task.
Second-order effects
Rival model providers face added pressure to demonstrate coding performance on demanding, externally recognizable evaluations rather than relying only on general benchmarks.
Buyers evaluating AI for difficult programming work may place more value on systems that can route hard tasks across specialized models, not just on a single model's score.
Third-order effects
If such results translate beyond contest settings, AI coding competition will increasingly center on orchestration—matching tasks to reasoning models and verification workflows—rather than a single flagship model.
Competitive-programming results can become useful capability signals, but the gap between solving bounded contest problems and reliable production software remains a key limit on how much they predict deployment value.
The trend: This is one data point in the shift from general-purpose chat models toward higher-cost reasoning systems and multi-model workflows for complex technical tasks.
3/n We competed with an ensemble of general-purpose reasoning models; we did not train any model specifically for the ICPC. We had both GPT-5 and an experimental reasoning model generating solutions, and the experimental reasoning model selecting which solutions to submit. GPT-5
Reasoning models (apparently without tool use) scored #1 (OpenAI) & tied for #2 (Google) in the International Collegiate Programming Contest Its been one year since reasoners were first announced, it is genuinely surprising how good they have gotten at hard problems, so quickly […
Yes, we have reached a turning point. OpenAI achieved a perfect score at the ICPC World Final 2025—the most important university programming competition—completing 12 out of 12 tasks in 5 hours, outperforming every human team. The AI team, which was not allowed to participate i…
GPT-5 solved the majority of problems, underscoring our achievement in providing one of the most intelligent and widely accessible AI models to the public. Just one year ago, our reasoning models were significantly less intelligent and far more expensive. Today, we've developed…
We used a simple yet powerful approach: We simultaneously generated multiple candidate solutions using GPT-5 and an internal experimental reasoning model, then used our experimental model to intelligently select the optimal solutions for submission. There was no complex strategy
1/5 In 2015, I won the ICPC World Finals as a member of the ITMO University team. It was the only time in Finals history when a team solved all the problems before the contest ended. [image]
Our general-purpose reasoning models solved all 12 problems at the 2025 International Collegiate Programming Contest (ICPC) World Finals, the world's top university programming competition which was enough for a 1st-place human ranking.
I'm excited to announce that an advanced version of Gemini Deep Think achieved gold-medal level performance at the 2025 ICPC World Finals, one of the world's most prestigious programming competitions! 🥇Learn more in our blog post: https://deepmind.google/... An inspiring moment f…
-ICPC 2025 (Programming Contest): Gold OpenAI / Google (first time outperforming humans) -IMO 2025 (Math olympiad): Gold OpenAI / Google This is the worst it will ever be. And the takeoff has begun.
Very excited to see our Gemini models getting better and better at coding! An advanced version of Gemini 2.5 Deep Think at the 2025 International Collegiate Programming Contest (ICPC) World Finals achieved gold-medal level performance! 🎉 https://deepmind.google/...
An advanced version of Gemini 2.5 Deep Think has achieved gold-medal level performance at the ICPC 2025 - one of the world's most prestigious programming contests. 🏅 Building on the model's success in math at the IMO, this marks another historic milestone for advanced AI. 🧵 [imag…
OpenAI and Google both win gold medals at the 2025 International Collegiate Programming Contest (ICPC) World Finals. OpenAI got 12 problems correct out of 12, beating the best human team (which got 11 out of 12). Google got 10 out of 12. Both competed under the same set of rules
🥇 HUGE NEWS: Google earned a gold medal at ICPC with an advanced version of Gemini Deep Think. ICPC: the world's most prestigious programming competition, requires abstract reasoning. Deep Think even solved one problem humans couldn't. 🚀 Software singularity is on schedule. [imag…
An exciting moment for AI in complex algorithmic reasoning & coding. Our new Gemini Advanced model achieved Gold at the ICPC, a programming contest close to my heart! Also the beginning of an great journey in this space. So proud of the amazing team: https://deepmind.google/... 1…
After Math comes Coding! @GoogleDeepMind Gemini 2.5 Deep Think achieved gold-medal level at the International Collegiate Programming Contest World Finals. - Achieved gold-medal level at the 2025 ICPC World Finals. - Correctly solved 10 out of 12 complex algorithmic problems. - [i…
Gemini achieves gold-medal level at the International Collegiate Programming Contest World Finals https://deepmind.google/... < great @GoogleDeepMind post on Gemini performance. Solved 10 of 12 problems, including one no university team solved!
An advanced version of our Gemini Deep Think model achieved 🏅 gold-medal level performance at the ICPC World Finals 2025. This marks a new milestone for AI in competitive programming. Read more: https://deepmind.google/... [image]
wow, this is exciting. both openai and google won gold medals at ICPC 2025. oai system solved 12/12 while gemini 2.5 deepthink solved 10/12. what is noteworthy is that no human team solved more than 11. [image]
Incredible milestone: an advanced version of Gemini 2.5 Deep Think achieved gold-medal performance at the ICPC World Finals, a top global programming competition, solving an impressive 10/12 problems. Such a profound leap in abstract problem-solving - congrats to @googledeepmind!
So Google Gemini just brute-forced the ICPC competition problems and managed to get a gold medal: arstechnica.com/google/2025/ .... But they also published the code produced by Gemini and well let me tell you, LET ME TELL YOU: LMFAOOOOO