|
|
|
|
Anthropic dropped a model that nearly matches its best for half the price. OpenAI plugged ChatGPT into your actual medical records. And Microsoft shipped an AI that finds security holes better than every frontier model on the market. Big week.
In This Issue:
- 💥 Big News: Claude Opus 5 doubles its predecessor, ChatGPT reads your lab results & Microsoft launches an AI cyber defender
- ✨ Weekly Spotlight: Generate ad creative, infographics & product shots free with Google's Nano Banana 2
- 🔥 Trending Tools: 6 new releases for agent monitoring, ad generation, deep research, code review & voice coding
- 🛠️ Deep Dives: We break down Originality.AI, Vexub & Travelnaut for content detection, video creation & trip planning
|
💥 Big News
Anthropic just released Claude Opus 5, and the pitch is simple: get close to the power of their top model, Fable 5, for half the cost. On coding and business tasks, Opus 5 more than doubles its predecessor's scores while keeping the same $5/$25 pricing. A new effort toggle lets you dial the model up for harder problems or down for faster, cheaper answers.
The same week, Anthropic also upgraded voice mode to run on Opus and Sonnet (instead of only lightweight Haiku), so you can now talk through complex work out loud and trigger connected tools like Gmail and Calendar mid-conversation. 
Why it matters:
- More than doubles Opus 4.8's benchmark scores at the same price, with a new effort toggle for cost control
- Available now on Claude Pro, Max, Team, Enterprise, Claude Code, and the API
- Voice mode now runs on Opus & Sonnet too, with tool access and 11 languages
Read more →
OpenAI just launched Health in ChatGPT for all US users. You can now connect Apple Health data and medical records from supported hospital systems, One Medical, and Function Health, then ask ChatGPT to compare lab results, summarize changes since your last visit, or spot patterns between sleep, activity, and your routine.
Over 300 million people already ask ChatGPT health questions every week. The difference now is that it can see your actual data instead of guessing from what you type. It works across Free, Go, Plus, and Pro plans on web and iOS, and your connected health data is not used to train models or target ads. 
Why it matters:
- Connects Apple Health, hospital records, One Medical & Function Health to your conversations
- Available to all US users 18+ on every ChatGPT plan, web and iOS
- Health data is encrypted, not used for model training or ads, and you control access per conversation
Read more →
Microsoft just unveiled MAI-Cyber-1-Flash, a purpose-built AI model that finds and helps fix security vulnerabilities in code. Paired with GPT-5.4 inside Microsoft's MDASH system, it scored 96% on CyberGym (the industry's primary security benchmark), beating models from Anthropic, Google, and OpenAI at half the cost.
Alongside it, Microsoft announced Project Perception, a platform that deploys teams of AI agents to monitor, patch, and close security gaps continuously. Perception enters public preview in Microsoft Defender on August 3. 
Why it matters:
- Scores 96% on CyberGym, 12 points above the next-best model, at half the cost
- Handles 90% of routine security work automatically, only escalating the hardest tasks
- Project Perception deploys AI agent teams for continuous threat monitoring, preview August 3
Read more →
|
✨ Weekly Spotlight
Google's Free Image Model Now Generates Production-Ready Assets
The Nano Banana 2 Quickstart Guide walks you through Google's latest image generation model, available inside any Google account at no cost. It covers one-shot prompt techniques for ad creative, infographics, event posters, and hero product shots, with eight ready-to-run starter prompts included.
Key Takeaways:
- Fast generation (images render in under 20 seconds, no queue or credits required)
- Precise editing (make selective changes to part of an image without redoing the whole thing)
- Accurate on-image text (renders readable text in generated images, useful for ads & infographics)
|
|
|
🛠️ Deep Dives
1. Originality.AI: A powerful tool to detect AI-generated content, check for plagiarism, and verify facts.
Why It's Worth Your Time:
- Detect AI content (accurately identifies text generated by various AI models)
- Check for plagiarism (compares your content against a vast database to ensure originality)
- Verify facts efficiently (cross-references information to boost content accuracy and credibility)
Pricing: Paid ($14.95/mo) for full AI detection, plagiarism, and fact-checking features.
2. Vexub: Transform text or audio into engaging, AI-generated videos with ease.
Why It's Worth Your Time:
- Convert text to video (turn written content into dynamic visuals quickly)
- Create from audio (transform spoken words into engaging video narratives)
- User-friendly interface (design videos easily, even without prior editing experience)
Pricing: Free trial available to test its text-to-video and audio-to-video capabilities.
3. Travelnaut: Unleash AI to tailor and simplify your travel planning journey.
Why It's Worth Your Time:
- Personalized trip planning (AI tailors suggestions to your preferences)
- Comprehensive travel info (access a vast database of destinations and activities)
- Simplify itinerary creation (effortlessly build and optimize your travel schedule)
Pricing: Free to use for all travel planning and information features.
|
⏭️ What's Next
If you skipped Opus because Fable was the only model worth switching to, this week changed the math. And if you have an iPhone with Apple Health data sitting unused, ChatGPT can now do something with it. Two things worth a ten minute try this weekend.
|
How did we do? |
|
|
|
No hay comentarios:
Publicar un comentario