• Everyday AI
  • Posts
  • Has Anthropic’s Claude lost its edge? What happened & can Claude recover?

Has Anthropic’s Claude lost its edge? What happened & can Claude recover?

OpenAI's new social network, Apple's AI boost with user data, GPT-4.5 phasing out, OpenAI's engineer AI agent and more!

Author

Everyday AI
April 15, 2025

Outsmart The Future

Today in Everyday AI
7 minute read

🎙 Daily Podcast Episode: Has Anthropic’s Claude lost its edge in AI? Discover why OpenAI and Google are pulling ahead and what it means for AI dominance. Give it a listen.

🕵️‍♂️ Fresh Finds: OpenAI’s new model naming, Google Classroom gets AI quiz feature and Cohere launches Embed 4. Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: OpenAI’s social network, Apple to boost AI features with user data and OpenAI phasing out GPT-4.5. For that and more, read on for Byte Sized News.

🧠 Learn & Leveraging AI: What happened to Anthropic’s Claude? We break down how it fell from the top and what it means for the AI landscape. Keep reading for that!

↩️ Don’t miss out: Did you miss our last newsletter? We talked about OpenAI's GPT-4.1 AI models, NVIDIA's U.S. AI supercomputers and NATO acquiring AI defense. Check it here!

 Has Anthropic’s Claude lost its edge? What happened & can Claude recover? 🏃

Seems like Google and OpenAI right now are just lapping the competition.

What happened to Anthropic's Claude, which was once the Internet's darling when it came to all things creative writing, problem solving and coding?

Can Claude catch up? Or will they quietly delve into obscurity?

Hot Take Tuesday is back with a vengeance.

Also on the pod today:

• Claude 3.7 and Industry Relevance 💼
• Claude User Experience and Rate Limits 🧑‍💻️
• Benchmark Performance vs. Competition 🏋

It’ll be worth your 51 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – Extrovert warms up LinkedIn prospects at scale, Aqua Voice is advanced speech to text for Mac and Windows and Potpie AI creates AI agents for your codebase in minutes.

OpenAI – Sam Altman has hinted at fixing OpenAI’s model naming this summer.

NVIDIA – NVIDIA will be opening Computex this year with a keynote from CEO Jensen Huang.

Google - Google Classroom is giving teachers an AI feature for quiz questions.

LLMs - Cohere has launched Embed 4, a multimodal embedding model that enables search for agentic AI.

Content Creation - Runway’s Gen 4 is now available on mobile.

Kling AI has just dropped its 2.0 phase for its video and photo generation.

Gamma has announced updates for Gamma 2.0, an AI design platform.

Trending in AI – Notion has released Notion Mail, an email client for Gmail.

AI Startups – Virtue AI has raised $30M in seed and Series A funding.

AI Tech - This prototype AI wearable uses machine learning to help blind people with navigation.

AI Governance – San Diego’s city council is considering banning AI software used to set rent pricing.

1. OpenAI Tests Social Network with AI-Powered Feed 👀

According to multiple sources, OpenAI is quietly developing an X-like social network prototype centered on ChatGPT’s image generation and social feed features. CEO Sam Altman has been seeking outside feedback as the company explores whether to launch this as a standalone app or integrate it into ChatGPT, which recently became the world’s most downloaded app.

This move could intensify competition with Elon Musk’s X and Meta’s upcoming AI-driven social features, highlighting a new battleground for real-time AI data and content creation.

2. Apple Leverages Synthetic Data to Boost AI Models Privacy 📊

In response to criticism over its AI features, Apple revealed a novel approach using synthetic data combined with differential privacy to enhance its AI models while keeping user data secure. By generating synthetic emails and polling opted-in devices with these anonymized snippets, Apple can better fine-tune its Genmoji and email summary tools without exposing real user content.

This method reflects a timely push for improved AI accuracy amid growing concerns around data privacy and model reliability.

3. OpenAI Phases Out GPT-4.5 API, Pushes GPT-4.1 as Successor 👋

OpenAI announced it will retire GPT-4.5 from its API by July 14, just months after its February debut, urging developers to switch to the freshly launched GPT-4.1. Despite GPT-4.5’s power and improved writing skills, its steep operating costs made it unsustainable for long-term API support.

GPT-4.1 promises similar or better performance at a fraction of the expense, signaling OpenAI’s focus on affordability without sacrificing quality.

4. OpenAI Unveils AI Engineer and $500B Data Center Push 👷

OpenAI is stepping up its game with A-SWE, an AI agent designed to fully replace software engineers by not only building apps but also handling testing and documentation, signaling a major shift from AI as a support tool to a standalone workforce.

Alongside this, the company is investing heavily in its own data center infrastructure through the $500 billion Stargate project, aiming to solve past computing power bottlenecks and rival cloud giants like AWS.

5. Gemini App Integrates Google Photos on Android 📸

The Gemini app has officially started rolling out integration with Google Photos for Android users in the US, following a preview last month, according to 9to5Google. This new feature lets users search their backed-up photos using natural language queries about faces, locations, dates, and even specific details within images—blurring the line between AI-powered photo management and personal assistant.

It opens fresh possibilities for professionals and creatives to quickly access visual content tied to projects or memories without manual sorting.

6. Adobe Bets Big on AI Video with Synthesia Investment 🤑

Adobe’s venture arm has invested in British AI startup Synthesia, signaling a strategic push into AI-driven video production just as the demand for video content surges globally. Synthesia, which boasts over 70% of Fortune 100 clients and recently surpassed $100 million in annual recurring revenue, offers a platform that creates lifelike AI avatars for faster, scalable video creation.

Despite strong growth, the startup isn’t focused on profits yet, prioritizing expansion and innovation in a competitive market that includes rivals like OpenAI’s new text-to-video tool.

🦾How You Can Leverage:

Claude was once in the top 3 AI models in the WORLD.

Now it doesn't even crack the top 10 in most benchmarks.

WTF happened?

A model that's less than TWO MONTHS OLD (Claude 3.7) already feels antiquated compared to Google and OpenAI's latest offerings.

And get this — even Google's SMALL language model (Gemma 3) now scores higher in human preference tests than Claude's flagship model.

The data is brutal. 

ChatGPT gets 3.9 BILLION monthly visits while Claude limps along with just 76 million.

That's not just losing the race. It’s not even being in the same competition anymore. So has Claude lost its edge in the LLM race? 

And will they ever catch up to Google and OpenAI? 

That’s what we tackled today on our #HotTakeTuesday show. 

Make sure to give it a watch/listen, then dive in below for our bigger takeaways. 

Here we gooooooooo 👇

1 – Playing the Wrong Game 🎮

Anthropic deserves genuine respect for prioritizing safety in their AI development approach.

Their research papers are top notch and their approach to safe AI is commendable. 

Buuuuuuuut they're fighting yesterday's war while the battlefield has completely transformed around them.

While obsessively chasing perfect research papers and safety protocols (which absolutely matter!), they’ve got straight up SLAPPED by Google. 

Whereas Google struggled in 2023 and the early parts of 2024, now they’re taking other AI Labs’ lunch money. 

And Anthropic was the first to cough up their change. 

With Google’s Gemini 2.0 Flash and Gemini 2.5 Pro releases, they’ve essentially erased Anthropic’s previous stranglehold for customers wanting dev-friendly power for complex tasks. 

Try this: 

Create a balanced scorecard for AI platforms across three essential criteria: capability, safety, and usability (rating each 1-10). 

This comprehensive approach reveals which platforms truly satisfy your complete needs rather than excelling in just one dimension while failing in others.

2 – The Brutal Benchmark Reality 😬

Anthropic’s brand spanking new Claude 3.7 Sonnet ranks EIGHTH in intelligence according to Artificial Analysis.

Not second. Not third. EIGHTH.

Creative writing? Not in the top 10 despite once being the gold standard.

(Sorry Twitter Trendsetters and AI Hipsters. Claude is NOT a top model when it comes to creative writing. Take it from someone that’s been getting paid to write for 20+ years. Generally, this just means people don’t know how to work with models.) 

And coding? 

ELO scores (where humans blindly choose the better output) show Claude isn't even a top-10 model ANYMORE.

The shocking truth? OpenAI's smaller GPT-4.1 mini model outperforms Claude 3.7 on coding benchmarks. A MINI model beating Claude's flagship offering!

(But sure… ignore the receipts.) 

Remember when we advised "use Claude for one-shot creative writing" like 16 months ago? That ship has sailed, hit an iceberg, and sunk to the ocean floor.

Try this: 

Run a genuinely blind test with five identical complex prompts across multiple AI platforms. 

Have someone remove all identifying information from the responses. Rate each without knowing which model generated it. The results will shatter your platform loyalty faster than Claude hits its rate limits.

Here’s our favorite easy way to run tests across multiple models at once. 

3 – From Innovation Leader to Feature Follower 🚶

Nine months ago, Claude represented 25% of our daily AI usage.

Today? 

A sad 5%.

Claude pioneered game-changing features like Artifacts — that brilliant system for rendering code and building dashboards directly in your browser. Then came Projects with custom instructions that had competitors scrambling.

Those glory days of Claude trailblazing? 

GONE.

Now they're frantically testing voice features that other platforms mastered ages ago while Google and OpenAI drop revolutionary models every few months.

In artificial intelligence, yesterday's innovator becomes tomorrow's imitator faster than you can say "rate limit exceeded." Which with Claude, is extremely fast. Lolz. 

Try this: 

If your team are heavy front-end AI users, meticulously track where you’re spending your time in AI chatbots, on what platform and doing what task. 

Make sure to have open lines of communication on model choice, outcomes, and time savings ROI. 

If you’re pushing front-end LLMs to their limit, we’re guessing — like us — you may be finding less and less utility out of Claude’s offerings. 

So….. what’s the #HotTakeTuesday takeaway? 

Hard. 

Yes, we’ll see them recover. We’ll see more impressive models and some worthwhile research and meaningful contributions like their MCP protocol. 

But will Anthropic’s Claude ever be mentioned again as a hands-down SOTA AI chatbot? 

As long as Google and OpenAI continue at their current pace, we don’t see that happening. 

Reply

or to participate.