AI News Recap: February 28, 2025
AI Giants Shift Alliances: OpenAI's New Partnerships, Google's Affordable Innovations, and Anthropic's Creative AI Ventures
It’s update o’clock—your AI highlights are here!
Hope you're doing well. As always, your recap of this week’s AI developments, industry news, and breakthroughs are below. Enjoy your day — and weekend!
🕖 Short on time? - A TL;DR section has been provided for you at the end of this post.
Catch up on Neural Buddies’ latest post …
Your Weekly Recap
OpenAI to Move Compute Needs from Microsoft to SoftBank
Google's AI Video Model Veo 2 Priced at 50 Cents/Second
Anthropic Benchmarks New AI Model with Pokémon
Google Unveils Free AI Coding Assistant with Generous Usage Limits
OpenAI Delays Deep Research Integration in API
OpenAI Offers In-Depth Research to ChatGPT Subscribers
Anthropic's Claude AI Plays Pokémon on Twitch at a Slow Pace
Anthropic's New AI Flagship Trained Cost-Effectively
Claude: Key Insights on Anthropic's AI
Google Gemini: Key Insights on New Generative AI Models
OpenAI Launches GPT-4.5 'Orion,' Its Largest AI Model
Top AI Models: Functions and Usage Guide
OpenAI to Move Compute Needs from Microsoft to SoftBank
Published: 2025-02-21
Key Points
OpenAI plans to shift compute needs from Microsoft to SoftBank by 2030.
SoftBank-backed Stargate project will provide three-quarters of OpenAI's data center capacity.
OpenAI's costs for running AI models are expected to grow significantly by 2030.
Source(s)
Google's AI Video Model Veo 2 Priced at 50 Cents/Second
Published: 2025-02-23
Key Points
Google's AI video model Veo 2 costs 50 cents per second.
Veo 2 can create video clips of two minutes or more.
Pricing compared to OpenAI's Sora model and Marvel movie budgets.
Source(s)
Anthropic Benchmarks New AI Model with Pokémon
Published: 2025-02-24
Key Points
Anthropic used Pokémon Red to test its new AI model, Claude 3.7 Sonnet.
Claude 3.7 Sonnet demonstrated "extended thinking" by successfully battling Pokémon gym leaders.
Games like Pokémon are often used as benchmarks for AI model capabilities.
Source(s)
Google Unveils Free AI Coding Assistant with Generous Usage Limits
Published: 2025-02-25
Key Points
Google launches Gemini Code Assist, a free AI coding assistant with high usage caps.
The tool offers 180,000 code completions a month, far exceeding GitHub Copilot's free plan.
Gemini Code Assist aims to attract developers early in their careers to Google's ecosystem.
Source(s)
OpenAI Delays Deep Research Integration in API
Published: 2025-02-25
Key Points
OpenAI delays API release of its deep research model to assess AI persuasion risks.
The deep research model is tested for persuasiveness but remains less effective than humans.
OpenAI's competitor, Perplexity, launches a similar deep research API product.
Source(s)
OpenAI Offers In-Depth Research to ChatGPT Subscribers
Published: 2025-02-25
Key Points
OpenAI expands deep research tool access to all paying ChatGPT users.
ChatGPT Plus, Team, Enterprise, and Edu users get 10 queries monthly.
Tech companies are competing to offer valuable AI research tools.
Source(s)
Anthropic's Claude AI Plays Pokémon on Twitch at a Slow Pace
Published: 2025-02-25
Key Points
Anthropic's Claude 3.7 Sonnet AI plays Pokémon Red on Twitch, showcasing AI capabilities.
The AI model uses reasoning to solve in-game puzzles, unlike its predecessor.
The livestream highlights AI's progress and challenges in navigating video games.
Source(s)
Anthropic's New AI Flagship Trained Cost-Effectively
Published: 2025-02-25
Key Points
Anthropic's Claude 3.7 Sonnet AI model cost 'a few tens of millions' to train.
Training costs for Claude 3.7 are relatively low compared to other AI models.
Future AI models may see increased costs due to advanced features and research.
Source(s)
Claude: Key Insights on Anthropic's AI
Published: 2025-02-26
Key Points
Anthropic's Claude AI models perform diverse tasks, including image captioning and email writing.
Claude 3.7 Sonnet is the most advanced model, offering hybrid reasoning capabilities.
Claude models are available via API with varied pricing based on model and usage.
Source(s)
Google Gemini: Key Insights on New Generative AI Models
Published: 2025-02-27
Key Points
Google Gemini is a suite of generative AI models offering multimodal capabilities beyond text.
Gemini models are integrated into Google services like Gmail, Docs, and Maps for enhanced functionality.
Google offers Gemini Advanced with premium features for complex tasks and corporate applications.
Source(s)
OpenAI Launches GPT-4.5 'Orion,' Its Largest AI Model
Published: 2025-02-27
Key Points
OpenAI launches GPT-4.5 'Orion,' its largest AI model, emphasizing research preview.
GPT-4.5 shows mixed performance, excelling in some benchmarks but lagging in others.
High operational costs and scaling limitations challenge GPT-4.5's long-term viability.
Source(s)
Top AI Models: Functions and Usage Guide
Published: 2025-02-28
Key Points
Overview of the latest AI models released since 2024, including their uses and strengths.
Highlights include OpenAI's GPT 4.5 'Orion' and Anthropic's Claude Sonnet 3.7.
Discusses accessibility and subscription requirements for various AI models.
Source(s)
TL;DR
Here’s a summary of the main points from all the articles …
OpenAI to Move Compute Needs from Microsoft to SoftBank: OpenAI shifts compute to SoftBank’s Stargate project by 2030, expecting rising AI model costs.
Google's AI Video Model Veo 2 Priced at 50 Cents/Second: Google’s Veo 2 video model costs 50 cents/sec, competes with OpenAI’s Sora.
Anthropic Benchmarks New AI Model with Pokémon: Claude 3.7 Sonnet tested with Pokémon Red, shows advanced reasoning.
Google Unveils Free AI Coding Assistant: Gemini Code Assist offers 180,000 free code completions/month, targets young developers.
OpenAI Delays Deep Research Integration in API: OpenAI delays deep research API release to study persuasion risks.
OpenAI Offers In-Depth Research to ChatGPT Subscribers: All paying ChatGPT users get 10 deep research queries/month.
Anthropic's Claude AI Plays Pokémon on Twitch: Claude 3.7 Sonnet livestreams Pokémon Red, highlights AI progress.
Anthropic's New AI Flagship Trained Cost-Effectively: Claude 3.7 Sonnet trained for tens of millions, cost-efficient.
Claude: Key Insights on Anthropic's AI: Claude 3.7 Sonnet excels in diverse tasks, hybrid reasoning, available via API.
Google Gemini: Key Insights on New Generative AI Models: Gemini suite enhances Google services, offers premium features.
OpenAI Launches GPT-4.5 'Orion': GPT-4.5 ‘Orion’ debuts as OpenAI’s largest model, faces high costs and mixed results.
Top AI Models: Functions and Usage Guide: Overview of 2024+ models like GPT-4.5 and Claude 3.7, with usage details.
This content is curated and compiled autonomously using a Python-driven AI system. While every effort is made to ensure accuracy, we welcome your feedback on any discrepancies you may notice.