Robots are doing go-karting now.

Robots & Art 🌈

🤯 A mysterious new AI model called “Ox Alpha” is reportedly BEATING Claude Fable 5 and GPT-5.6 Sol at coding, and nobody knows who built it.

The model just appeared on OpenRouter with a 1 million-token context window, multimodal capabilities, and no disclosed owner.

Speculation is exploding, with some online sleuths pointing to Google, while others believe it could be a new model from Chinese AI lab Z .ai.

Adding to the mystery, Ox Alpha’s tokenizer reportedly appears identical to GLM’s, and the previous four anonymous AI model drops were all eventually claimed by Chinese labs.

@aipost 🏴

Google's quantum chip cracked a problem in 5 that would take the world's fastest supercomputer tens of BILLIONS of years.

CEO Sundar Pichai says it proves quantum error correction now works at scale, putting practically useful quantum computers 5 to 10 years away.

@aipost 🏴

Sam Altman’s sister on X

@aipost 🏴

❗️The rise of AI writing is leaving a stylistic footprint on the web.

Pewresearch:

"And as more AI-generated content appears online, these and other AI “tells” have become more common across webpages. Comparing the internet of today to a snapshot from 2023:

• Em dashes appear about twice as frequently.
• Oxford commas see a 63% increase.
• Certain words that AI models like to use (such as “delve,” “interplay” or “testament”) have more than doubled in usage.
• Using “negative parallelism” to structure a comparison (“it’s not just X, it’s Y”) has nearly tripled, although this is still fairly rare overall."

@aipost 🏴

This is very impressive.

This requires autonomous real-time planning and action in response to a fast-moving target and dynamic environment.

@aipost 🏴

🤖 ChatGPT can now send iMessages

ChatGPT is getting a serious upgrade on Mac: it can now connect to Messages and actually interact with your conversations.

The new Messages plugin can find chats, catch you up on messages you missed, draft replies, and even send messages for you.

It can also connect the dots between your messages and calendar, like spotting someone’s birthday in a chat, adding it to your calendar, or finding an open evening for dinner.

ChatGPT can flag conversations that look like spam, turning your messages into something it can actually help manage instead of just summarize.

@aipost 🏴

🗣 Dave Chappelle says no one needs to think anymore because of AI:

“Thinking is for poor people”

“I was talking to a friend of mine and we were laughing about how no one really would know what to teach a kid now. What skill sets are going to matter 10 or 20 years from now as the world becomes automated”

“We decided your character is your destiny. If you’re a good person that’s all you can ask from anybody and hope for the best”

“I don’t know if anyone needs to think anymore. It’s AI, quantum computing, it seems like thinking is for poor people”.

@aipost 🏴

Patients are now being advised to consult artificial intelligence tools prior to medical appointments to gain a clearer understanding of their health issues and possible treatments. Medical professionals are also expected to integrate AI into their workflow, using it as an additional step to verify treatment plans.

Following the introduction of advanced AI platforms like ChatGPT, both patient and clinician usage of the technology increased significantly. Health technology experts continue to examine strategies for addressing challenges related to the integration of AI in clinical practice, with a focus on ensuring reliable and transparent decision-making.

📰 @aipost

Nvidia demonstrates up to five times better cost efficiency compared to AMD at a rate of 150 tokens per second per user, according to the latest SemiAnalysis report.

Data indicates that with current AMD software, Nvidia's performance lead is significant. Even if AMD hardware were provided at no cost, while still accounting for expenses such as datacenter hosting and power, Nvidia would continue to deliver a lower cost per token.

This comparison is based on GLM 5.3 SGLang benchmarks and focuses on operational costs under real-world usage conditions.

📰 @aipost

OpenAI has announced a reduction in its GPT-5.6 Sol API pricing, effective immediately. Input token costs have dropped by 20%, now set at $4 per million. Output token pricing has been reduced by 33%, with the new rate at $20 per million.

These favorable rates are guaranteed at least until November 21. The price adjustments follow closely on the anticipated Anthropic IPO, indicating ongoing competition in the AI sector.

No links or promotions are included in this update.

📰 @aipost

🚀 SpaceX is putting AI data centers in orbit

SpaceX is reportedly taking NVIDIA’s Vera Rubin architecture to space with its new Starmind satellites.

Each satellite could pack 72 Rubin GPUs + 36 Vera CPUs into a custom SpaceX-designed rack that’s lighter, denser and simpler than NVIDIA’s traditional NVL72 setup.

And these aren’t just satellites with some AI chips onboard. They’re essentially datacenters in orbit, solar-powered, running datacenter-class compute, with results beamed back to Earth through Starlink laser links.

SpaceX says the same radical rack design will also be used on the ground, while Elon says the company is “exclusive to NVIDIA.”

SpaceX is talking about scaling Grok’s ground compute from 2 GW to 10 GW, with a potential fleet of up to 1 million satellites. First Starmind launch in Q4 2027.

@aipost 🏴

A humanoid robot from Team Tianjiao achieved a 7.97-metre long jump, claiming the gold medal at the 2nd World Humanoid Robot Games in Beijing.

This performance is only 0.98 metres short of Mike Powell’s men’s world record of 8.95 metres.

In 2025, the best recorded jump for a humanoid robot was 1.25 metres.

📰 @aipost

A recent study has found that language models evaluate a user's expertise before generating responses. This internal assessment impacts both the detail in their answers and the likelihood of engaging with the user.

Researchers observed that models apply a single framework when gauging user competence, which then shapes the complexity of the information provided.

The paper concludes that while this computed estimate directly influences model behavior, observed demographic patterns linked to it do not consistently result in discriminatory output.

The research, titled “Who Do Language Models Think Is Competent? A Mechanistic Analysis of Occupational Bias,” focuses on understanding occupational bias in language model interactions.

📰 @aipost

❗️Your boss wants AI to read your slack messages

Executives are increasingly encouraging employees to use public Slack channels instead of private DMs so AI agents can access more context, analyze conversations, summarize meetings, build presentations and even help make decisions.

At Zapier, CEO Wade Foster encourages employees to default to public messaging. He recently used an AI agent to turn two months of Slack conversations about a customer into a presentation, during the customer call itself.

ElevenLabs has taken it even further: internal communication happens on Slack, while an AI agent turns updates from roughly 100 public channels into a podcast employees can listen to. The more information AI can access, the more useful the agents become.

Private chats are often where employees brainstorm, vent and discuss sensitive ideas freely. Making everything AI-readable could change how people communicate at work and raise some very uncomfortable privacy questions.

Source.

@aipost 🏴

🗣“A superintelligence would be a million times smarter than us.”

Roman Yampolskiy, associate professor of computer science at the University of Louisville, warns that once AI surpasses humans at everything, we may no longer understand what it is doing.

The real shift is not smarter tools. It is losing the ability to predict and control the systems shaping our world. Open intelligence only matters if humans remain in the loop.

@aipost 🏴