- The Rundown AI
- Posts
- OpenAI's GPT-4.5 is here
OpenAI's GPT-4.5 is here
PLUS: China's Tencent drops new 'fast-thinking' model
Rowan Cheung
February 28, 2025
Good morning, AI enthusiasts. OpenAI just dropped its largest model yet — but instead of crushing benchmarks, GPT-4.5 is doubling down on *feeling more human*.
With Sam Altman calling it "the first model that feels like talking to a thoughtful person," will users embrace this emotional evolution or scoff at the lack of raw performance gains they've come to expect?
In today’s AI rundown:
OpenAI’s GPT-4.5 with emotional intelligence
Tencent’s new ‘fast-thinking’ model
Building a smart calendar management system
Ideogram eyeing speed boost with new model
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
OPENAI
Image source: OpenAI
The Rundown: OpenAI just released GPT-4.5 (code-named Orion), the company’s largest model to date — which uses unsupervised learning instead of reasoning to achieve deeper world knowledge and improved emotional intelligence.
The details:
OpenAI says GPT 4.5 delivers a more natural conversational experience, with an improved understanding of human intent and greater emotional intelligence.
The model hallucinates less and delivers more accurate answers than previous versions, with testers liking it for pro tasks, creative work, and everyday queries.
It isn't a step up from previous models on math or science but does surpass o3-mini and o1 on SWE-Lancer, OpenAI’s new freelance coding task benchmark.
Only Pro users and developers on paid plans can access GPT-4.5 immediately, with Plus and Team users gaining access next week.
Notably, the API price of the model has been kept shockingly high at $75/$150 per million input/output tokens. For reference, GPT-4o costs just $2.50/$10.
Why it matters: While the benchmarks and pricing might leave some disappointed, 4.5 seems like more of a ‘vibe’ personality upgrade than a major step up. With high costs and fewer improvements than users have come to expect, this might also be the last stop both practically and acceleration-wise in non-reasoning model development.
TOGETHER WITH TURING
The Rundown: What happened at NeurIPS 2024? This recap breaks down the biggest AI advancements from leaders at Oracle, IBM, and Turing so you can stay ahead without the noise.
Gain insights on:
Multi-agent reasoning: How AI systems are learning to collaborate
Real-time adaptation: Smarter, more responsive models in action
Frontier scaling: Pushing the limits of training and optimization
TENCENT
Image source: Tencent
The Rundown: Chinese giant Tencent just released Hunyuan Turbo S, a new ‘fast-thinking’ AI designed for instant responses rather than deep reasoning — achieving 2x the speed while matching the performance of leading models on key benchmarks.
The details:
Turbo S matches models like DeepSeek V3, GPT-4o, and 3.5 Sonnet across knowledge, mathematics, and reasoning despite a focus on speed.
Tencent has significantly lowered the price of the new model, making it a fraction of the cost of the previous generation.
The company is also preparing to launch a complementary T1 reasoning model with "deep thinking," positioning the two models for different use cases.
The release comes amid increasing AI competition from China, with DeepSeek nearing a new launch and Alibaba debuting QwQ-Max for reasoning this week.
Why it matters: It wasn’t long ago that reasoning models were the new shiny toy, and now we have a ‘fast-thinking’ vs. ‘slow-thinking’ divide. With DeepSeek’s R1 shining a massive global spotlight on Chinese AI, rival labs are quickly rushing to one-up the industry darling — and U.S. chip restrictions don’t seem to be slowing anything down.
AI TRAINING
The Rundown: In this tutorial, you will learn how to create a personalized AI calendar assistant that can check your schedule and create events using n8n's workflow automation and AI capabilities.
Visit n8n and create a new workflow.
Add a chat trigger as your entry point.
Set up an AI Agent with calendar tools (Get Events and Create Event) using AI expressions, such as .
Add system instructions for the AI agent node to guide the AI's behavior.
Pro tip: Test your assistant with natural language prompts like "Find a one-hour slot tomorrow in my calendar and schedule a research meeting with Alex."
Check out our recent workshop showing how to create AI agents with n8n here.
PRESENTED BY AUGMENT CODE
The Rundown: Augment Code isn’t your average AI assistant — it’s purpose-built for serious software engineers handling massive codebases. Backed by Eric Schmidt and trusted by top teams like Webflow, Lemonade, and DDN, Augment helps take productivity to the next level.
With Augment Code, you can:
Instantly integrate into VS Code, JetBrains, or Vim to stay in the zone
Leverage full-codebase context in under 200ms to keep your workflow fast
Handle tens of thousands of files or millions of lines effortlessly
IDEOGRAM
Image source: Ideogram / The Rundown
The Rundown: Ideogram launched its 2a model, a major update to the text-to-image platform that significantly reduces generation time and cost while maintaining high-quality outputs—with optimizations for graphic design and photorealistic generations.
The details:
2a generates image outputs in just 10 seconds, with an even faster ‘2a Turbo’ option delivering results at twice the speed.
The new model excels at graphic design and text generation, with the ability to create content like homepages, movie posters, and advertisements.
It is also optimized for photorealism and is priced at 50% less than Ideogram 2.0 for both API and web use.
Users can access it now via Ideogram’s web platform, API, or through applications like Freepik, Poe, and Gamma.
Why it matters: While many image-generation models still struggle with basic spelling, Ideogram’s latest release is taking graphic design and text generation to absolutely mindblowing new levels. With these new abilities, another complete step change in creative workflows is about to be unlocked.
QUICK HITS
💬 Scribe - ElevenLabs’ new SOTA speech-to-text model
🪨 Granite 3.2 - IBM’s compact open models for enterprise use
🗣️ Octave TTS - Generate AI voices with emotional delivery
🧑🔬 Deep Review - AI co-scientist for literature reviews
Pika Labs released its new 2.2 model, featuring upgraded quality, 10-second, 1080p resolution generations, and new transition and transformation capabilities.
Meta is reportedly planning to release a standalone Meta AI app in Q2 with potential paid subscription options similar to OpenAI's model.
Figure is pushing up its timeline to bring its humanoid robots into the home, beginning Alpha testing this year thanks to improvements from its recently revealed Helix AI.
Microsoft rolled out new updates to Copilot, including a dedicated MacOS app, the ability to upload PDF or text files, and UI improvements.
Meta introduced Aria Gen 2 glasses with advanced sensors, on-device AI processing, and an all-day battery for research in machine perception, contextual AI, and robotics.
You Labs unveiled ARI, a research agent capable of analyzing up to 400 sources and generating professional reports with charts, citations, and visuals in under 5 minutes.
COMMUNITY
Join our workshop today at 4 PM EST to learn how to take full advantage of Claude 3.7 Sonnet’s frontier reasoning capabilities, along with advanced prompting strategies with Dr. Alvaro Cintas, The Rundown’s AI professor.
RSVP here. Not a member yet? Join The Rundown University on a 14-day free trial.
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
That's it for today!Before you go we’d love to know what you thought of today's newsletter to help us improve The Rundown experience for you. |
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Reply