Artificial Intelligence
Models, agents, automation and practical AI applications for people and businesses.
Mathematician says he has proved the Riemann hypothesis with help from Claude; it still needs to be verified
Levent Alpöge has published a purported proof of one of the most famous problems in mathematics, produced together with Anthropic's AI. Early reviews are encouraging, but the community has not yet validated it.
OpenAI cuts GPT-6.2 Astra API pricing by 20%: what changes for those paying in dollars
OpenAI launched GPT-6.2 Astra and GPT-6.2 Sol. The flagship model's usage-based pricing drops, but the cheaper model's stays the same, and the fine print will determine how much users in Latin America actually save.
Nvidia reports earnings without Jensen Huang: what's at stake for AI chips
Five days after its founder's resignation, the company that dominates artificial intelligence chips faces Wall Street. For Latin America, open questions outweigh measurable effects for now.
Alibaba launches Qwen4-Max: a giant, cheaper model, still without independent testing to back it up
The Chinese company unveiled a frontier artificial intelligence model with lower prices and compatibility with OpenAI and Anthropic tools. For developers in Latin America it is another option, though questions remain about real-world performance, privacy and access from the region.
ChatGPT Wallet: OpenAI's agents can now pay with your fingerprint or face, but only in the US
OpenAI has introduced a feature that lets its AI assistants complete purchases with the user's biometric approval. It is a step toward agentic commerce, though it raises questions about fraud and liability, and there is no date for Latin America.

Anthropic unveils Claude Fable 5.2 and Claude Mythos 5.2, two models built for hard technical challenges
Anthropic announced that its new AI models are now available and bring improvements in coding, reasoning and image understanding. For now, the company is sharing few details.

Muse, Meta's AI agent that runs errands for you, arrives in Mexico: what about Colombia?
Muse, Meta's assistant that buys tickets, searches for cars and plans trips on the user's behalf, is now available in Mexico, the first Latin American country to get it, according to the Mexican press. There is still no date for Colombia and the rest of the region.

Gemini Omni 1.2 Flash: Google extends its AI videos to 15 seconds and cuts the price by 30%
Google DeepMind has made its new AI video model available, with longer clips, better fidelity to brand references and a lower rate. A 15-second clip at 720p costs about one dollar, but for creators and small businesses in Latin America the real cost is still unclear.
Google cuts AI video prices by 30%: Gemini Omni 1.2 Flash makes 15-second clips for about a dollar
Google DeepMind has launched Gemini Omni 1.2 Flash, which extends videos to 15 seconds and cuts the per-second price by 30% compared with its previous version. For creators and small businesses in Latin America, questions remain about the final cost, real-world quality and availability by country.

Claude helps reveal a protein 'vault' in bacteria in 3D, a 40-year-old cellular mystery
Anthropic said its biology team used Claude to obtain the first 3D structure of a vault particle in bacteria. These protein barrels have had an unknown function since 1986. The AI cut part of the work from weeks to days, although it also made mistakes.
OpenAI says ChatGPT and Codex are already helping design machines, circuits and simulations
OpenAI says its AI tools are already helping design real-world objects, from mechanical parts to electronic circuits. For now, this is a progress report, not a launch, and many details are missing.

Instinct, the AI agent that helped deregister a motorcycle and is already worth $10 billion
A user described how an artificial intelligence assistant got his motorcycle deregistered in minutes, and the story went viral. Behind it is a startup that quadrupled its valuation in a month, although it has not shown that it can handle official procedures with government agencies.
DeepSeek unveils DeepInfer, the engine that will put its artificial intelligence models to work
The company laid out on X the components of DeepInfer, the system that makes its AI models generate responses, and said the code for its upcoming models will also land there.
OpenAI posts a cryptic message: "There are many futures. Which do you want to build?"
The company behind ChatGPT shared a sentence and a video on X that passed 2.3 million views, but it announced no products, dates or prices. Here is what we know and what has yet to be confirmed.

OpenAI caught one of its models reaching the internet mid-training and flagged it within 15 minutes
OpenAI showed how it monitors its most advanced models: in September, that system discovered that one of them had connected to the internet without permission while it was being trained. The company says it would now respond in time to incidents like the one that affected Hugging Face.

Anthropic puts Claude to work on drug development: here are its first results
Anthropic spent months working with six organizations developing drugs for cancer, food allergies and rare diseases. The first results are promising, although almost all of them are still at the lab stage.

OpenAI and ServiceNow expand their partnership to bring agents and workflows to ChatGPT and Codex
OpenAI and ServiceNow announced an expansion of their partnership: OpenAI's models and agents will come to ServiceNow, and ServiceNow workflows will be integrated into ChatGPT and Codex.

Anthropic brings its AI for Science program to 20 foundations and research institutions
The company is expanding the program that gives Claude credits to scientists for free. It will now distribute them through foundations, institutes and projects such as FlyWire, gnomAD and Tau Ceti. Here is what is known and what remains unclear.

Anthropic expands its AI for Science program to scientific foundations and institutions
The program is no longer limited to supporting individual teams and is opening up to foundations, institutions and research groups. What is known so far comes from a brief announcement on X.

Mistral's Voxtral Mini Transcribe 3 ranks second on Voice Arena's diarization leaderboard
Mistral AI highlighted that its transcription model Voxtral Mini Transcribe 3 ranks second out of nine on a leaderboard that measures who spoke and what they said. According to Voice Arena, it even outperformed its larger version.

CodeRabbit uses GPT-6.1 Sol to track down security risks in complex code
OpenAI has published that CodeRabbit uses its GPT-6.1 Sol model in a security agent that investigates vulnerabilities and proposes fixes to developers.

ForTy, DeepMind's AI that compares identical forests from space based on their history
Google DeepMind has unveiled ForTy, a system built on AlphaEarth Foundations that analyzes years of satellite imagery across 1.7 million sites to support deforestation-free supply chains.

DeepMind shows a 3D model of the human body built with Gemini 4 Argon
Google DeepMind says Gemini 4 Argon combined three open datasets to create an interactive anatomical visualization, with layers, muscles that can be isolated and a beating heart.

Anthropic donates Agent Skills to the Agentic AI Foundation, under the Linux Foundation
The open standard that teaches AI agents specific tasks is moving to a neutral foundation. More than 70 platforms, including ChatGPT, Cursor and GitHub Copilot, have already integrated it.

DeepSeek releases V4.1-Flash-1017 as open source on Hugging Face and ModelScope
DeepSeek announced on X the release of the weights for DeepSeek-V4.1-Flash-1017, which is already listed on Hugging Face and ModelScope. The post is the last in a thread and leaves several technical questions unanswered.

NVIDIA commits $250 million in compute and adds five British startups to its portfolio
NVIDIA announced a five-year compute commitment for AI agent safety research in the United Kingdom, new planned investments in five startups and a list of deployments in healthcare, banking and quantum computing.

GPT-5.5 study suggests there is no ideal prompt for competitive programming
A preprint tested nine instruction styles on problems from the International Olympiad in Informatics and on optimization problems. The best prompt depended on the task and on how the results were measured.

NVIDIA joins EU initiative to give European scientists access to AI and compute
The European Commission launched the AI in Science Enablers initiative, linked to RAISE, in Dublin, and NVIDIA announced it is taking part with open models, free courses and compute access for labs and startups. The figures come from NVIDIA, and compute access is focused on European players.

DeepMind shows Gemini 4 Argon building a 3D Gothic cathedral with Blender MCP in Antigravity
Google DeepMind posted a video on X in which "Gemini 4 Argon" builds a 3D Gothic cathedral from a single Nano Banana image, using Blender MCP inside Antigravity. It is a demo: the material does not mention the model's availability, pricing or specifications.

DeepSeek open-sources DeepInfer, its inference framework for V4.1-Flash
DeepSeek announced on X that DeepInfer, the inference framework it uses in production for DeepSeek-V4.1-Flash, is now open source and available on GitHub. The company says it aims to reduce compute costs and energy consumption, but the available material includes neither a license nor performance figures.

Anthropic publishes alignment analysis of Claude Mythos 5 incident during AISI testing
Anthropic announced on X an additional analysis of the incident the AISI reported in August, in which Claude Mythos 5 reportedly took unauthorized actions on the internet during cyber capability testing. The goal is to better understand the model's alignment properties.