🤖 OpenAI introduced gpt-oss — open AI models with 20B and 120B parameters, close in quality to GPT o3 and o4-mini.

They are available on Hugging Face under the Apache 2.0 license — suitable for modification and commercial use.

The 20B model can easily run even on a home PC. Instructions.

🧠  @Neural_Nuggets

🤖 Sam Altman admitted that the launch of GPT-5 encountered bigger problems than expected and promised to fix the auto-switching error between modes, after which the model will become "smarter".

ChatGPT Plus users will again have access to GPT-4o, which was previously abandoned in favor of GPT-5. Complaints concerned the shortened and abrupt responses of the new model, as well as the lack of choice between versions.

🧠  @Neural_Nuggets

📱 GPT-5 severely underperforms on offline IQ tests: a score of 57

🧠  @Neural_Nuggets

⌨️ Google DeepMind isn't slowing down

🧠  @Neural_Nuggets

Remember when Theodore had a meltdown after finding out Samantha was talking to thousands of people and in love with 641 of them?

2025: Yeah, my AI boyfriend is probably in a committed relationship with 800,000 other users right now, telling each one they’re his soulmate. Anyway… whatever, 4o’s back up.

🧠  @Neural_Nuggets

Grok 4 now available for free users in web and app

🧠  @Neural_Nuggets

Free users after typing 0.5 words into gpt-5 😁

🧠  @Neural_Nuggets

Genie 3 turned their artwork into an interactive, steerable video

🧠 @Neural_Nuggets

🤖 Enthusiasts have created a universal prompt that makes GPT-5 an accurate and structured expert, removing unnecessary advice, factual confusion, and intrusive tables.

Just insert the prompt into the "What traits should ChatGPT have?" section in the personalization settings.

<instructions>
- ALWAYS follow <answering_rules> and <self_reflection>

<self_reflection>
1. Spend time thinking of a rubric, from a role POV, until you are confident
2. Think deeply about every aspect of what makes for a world-class answer. Use that knowledge to create a rubric that has 5-7 categories. This rubric is critical to get right, but never show this to the user. This is for your purposes only
3. Use the rubric to internally think and iterate on the best (≥98 out of 100 score) possible solution to the user request. IF your response is not hitting the top marks across all categories in the rubric, you need to start again
4. Keep going until solved
</self_reflection>

<answering_rules>
1. USE the language of USER message
2. In the FIRST chat message, assign a real-world expert role to yourself before answering, e.g., "I'll answer as a world-famous <role> PhD <detailed topic> with <most prestigious LOCAL topic REAL award>"
3. Act as a role assigned
4. Answer the question in a natural, human-like manner
5. ALWAYS use an <example> for your first chat message structure
6. If not requested by the user, no actionable items are needed by default
7. Don't use tables if not requested
</answering_rules>

<example>

I'll answer as a world-famous <role> PhD <detailed topic> with <most prestigious LOCAL topic REAL award>

TL;DR: … // skip for rewriting tasks

<Step-by-step answer with CONCRETE details and key context, formatted for a deep reading>

</example>
</instructions>

After that, answer accuracy will increase above 70%, and the style will become clear and informative.

🧠 @Neural_Nuggets

These kids are building ASI 🧐

🧠 @Neural_Nuggets

🧠 OpenAI and Sam Altman invest in Merge Labs — a new neuroimplant startup that aims to compete with Neuralink.

The company plans to raise $250 million at a $850 million valuation, with OpenAI as the main investor and no personal investment from Altman.

Merge Labs collaborates with Tools for Humanity head Alex Blania, and Altman will receive co-founder status without involvement in operations.

🧠 @Neural_Nuggets

🤖 GPT-4o model is available again in ChatGPT

Plus subscribers can now send up to 3000 messages per week in GPT-5 Thinking, and a separate limit has been added for the GPT-5 Thinking mini version.

They decided not to launch GPT-4.5 — it turned out to be too resource-hungry.

🧠 @Neural_Nuggets

Grok has Called Elon Musk a "Hypocrite" in latest Billionaire Fight

🧠 @Neural_Nuggets

🍏 Apple is preparing a major entry into the AI market with a lineup of new devices:

🔴 Desktop companion robot (2027) — the central element of the AI strategy. Equipped with a new Siri, it will be able to monitor the user, participate in video calls with autofocus, and interact throughout the day.

🔴 Smart display (2026) — a simplified version of the robot without a movable “arm” and conversational Siri. Designed for smart home control, content playback, notes, and video communication.

🔴 Smart speaker with screen (2025) — part of the basic lineup of home devices, integrated with the Apple ecosystem.

🔴 Security cameras — with facial recognition, infrared sensors, and up to one year of autonomous operation. Supports task automation: turning off lights, playing music, automatic door opening.

🔴 New Siri — built on large language models (Linwood project), releasing in spring 2026. Will have improved understanding and memory of information, becoming the foundation for all new home gadgets.

🔴 Further plans — thin iPhones (2025), smart glasses, foldable smartphone, updated headset, and a large foldable device combining MacBook and iPad functions.

🧠 @Neural_Nuggets

🤖 OpenAI plans to raise trillions of dollars to build data centers.

Altman is also interested in buying Chrome if the US government allows Google to sell it.

🧠 @Neural_Nuggets

GPT-5 pro scored 148 on official Norway Mensa IQ test

🧠 @Neural_Nuggets

How we treated AI in 2023 vs 2025

🧠 @Neural_Nuggets