Updates and Insights for Today

1. Introducing GPT-4.1 in the API

2. ChatGPT Adds Built-In Image Library for Users

3. Explainable AI Makes Ship Navigation Safer

4. The latest in AI tech

5. AI Tutorials: Run an AI model offline on your computer.

6. AI tools to checkout

AI News

Introducing GPT-4.1 in the API

OpenAI has officially launched GPT-4.1, along with its smaller variants, GPT-4.1 Mini and GPT-4.1 Nano, as of April 14, 2025. These models are accessible exclusively through the OpenAI API and are not yet integrated into ChatGPT .

Key Features of GPT-4.1

• Expanded Context Window: All three models support up to 1 million tokens, a substantial increase from the 128,000-token limit of GPT-4o. This enhancement allows for more effective processing of extensive codebases and lengthy documents .

• Enhanced Coding Capabilities: GPT-4.1 demonstrates a 21.4% absolute improvement over GPT-4o and a 26.6% absolute improvement over GPT-4.5 on the SWE-bench Verified benchmark, establishing it as a leading model for coding tasks .

• Improved Instruction Following: The model achieves a 10.5% absolute increase over GPT-4o on the MultiChallenge benchmark, indicating better adherence to complex instructions .

• Advanced Long-Context Comprehension: GPT-4.1 sets a new state-of-the-art result by scoring 72.0% on the Video-MME benchmark for long videos without subtitles, a 6.7% absolute improvement over GPT-4o .

• Cost Efficiency: GPT-4.1 is 26% less expensive than GPT-4o for median queries. The Nano variant is the most affordable, priced at $0.10 per million input tokens and $0.40 per million output tokens .

Performance of Mini and Nano Variants

• GPT-4.1 Mini: This version matches or exceeds GPT-4o in intelligence evaluations while reducing latency by nearly half and lowering costs by 83% .

• GPT-4.1 Nano: Designed for tasks requiring low latency, such as classification or autocompletion, Nano is OpenAI’s fastest and most cost-effective model to date. It maintains the 1 million token context window and delivers exceptional performance at a small size .

Availability and Transition Plans

• GPT-4.1 and its variants are currently available exclusively via the OpenAI API.

• OpenAI plans to deprecate the GPT-4.5 Preview in the API by July 14, 2025, as GPT-4.1 offers improved or similar performance at a lower cost and latency  .

Prompting Guidance

To maximize the benefits of GPT-4.1, OpenAI recommends providing clear and specific instructions, including context examples, to guide the model effectively. GPT-4.1 is trained to follow instructions more closely and literally than its predecessors, which tended to infer intent more liberally

📌 Read More: OpenAI

Our Partner Today

Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

1. Join the Superhuman AI newsletter – read by 1M+ people at top companies

2. Master AI tools, tutorials, and news in just 3 minutes a day

3. Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

Sign up and start learning AI

ChatGPT Adds Built-In Image Library for Users

Quick Summary:

OpenAI has introduced an integrated image library in ChatGPT, giving users easy access to their AI-generated images. The feature is available on both mobile and desktop.

Key Insights:

• New “Library” section shows past generated images.

• Images are organized in a grid view for quick access.

• Rollout includes Free, Plus, and Pro users.

• Available now on iOS; web version coming soon.

Why It’s Relevant:

This move strengthens ChatGPT’s creative toolset. It simplifies workflows for creators and professionals using DALL·E-generated images. With growing demand for visual content, the feature enhances usability and offers a streamlined creative experience directly within the app.

📌 Read More: The Verge – Meta AI Memory

Explainable AI Makes Ship Navigation Safer

Quick Summary:

Japanese researchers have developed an explainable AI model for ship navigation, enhancing maritime safety by making decisions transparent and understandable to human crews.

Key Insights:

• The AI quantifies collision risk and explains its decisions.

• It communicates both rationale and intent of actions.

• Designed to improve human-AI trust in sea navigation.

• Developed by Osaka Metropolitan University.

Why It’s Relevant:

Maritime AI is advancing from automation to explainability. This new model could be a cornerstone for safe, autonomous vessels. By making AI choices clear to human operators, it addresses one of the key barriers to AI trust—especially in high-risk environments like busy sea routes.

📌 Read More: The Verge – Meta AI Memory

 AI Tutorials

I Built Jarvis using AI! The Ultimate AI Voice Agent with No Code

Quick Summary:

This video tutorial walks viewers through the process of creating a fully functional AI voice assistant—similar to Jarvis from Iron Man. The assistant can manage emails, schedule calendar events, handle to-do lists, and even conduct online research. It uses tools like n8n, ElevenLabs, Notion, Google Calendar, and OpenAI APIs.

Key Insights:

• The assistant integrates ElevenLabs for voice input and output, and n8n to orchestrate workflows.

• Core capabilities include sending emails, managing a Notion-based to-do list, updating a Google Calendar, and performing web research.

• The system uses JSON templates and webhooks for modular task handling.

• A CRM in Notion stores contact data, which the assistant uses to personalize tasks like sending emails.

What Can I Learn?

• How to link voice input with task execution using webhooks.

• How to automate email and calendar management with Google APIs.

• How to use Notion as a dynamic task manager and CRM.

• How to configure AI agents to take context-aware actions using n8n workflows.

Which Benefits Do I Get?

• You gain a fully automated productivity system powered by voice.

• You reduce manual scheduling, emailing, and organizing tasks.

• You get a customizable system fit for solopreneurs, startups, or tech-savvy professionals.

• You learn how to integrate multiple modern tools without writing code.

Why It’s Relevant:

Voice assistants are quickly becoming essential tools in personal productivity and business automation. This DIY project demonstrates how to build a robust AI assistant using open-source tools and APIs—perfect for solopreneurs, developers, and productivity enthusiasts. It also highlights the growing importance of integrating voice AI with everyday platforms like Notion and Google Workspace.

By leveraging systems like ElevenLabs and n8n, users can build voice-driven assistants that don’t just answer questions—they act on them, making automation more natural and accessible.

Here is the full Video Tutorial 👉 Click Here

The latest in AI tech

Hugging Face Acquires Robotics Startup

Hugging Face has acquired Pollen Robotics, the creators of the open-source humanoid robot “Reachy 2.” The company plans to distribute the robot and make it easier for developers to build AI-powered robotic applications. This move signals Hugging Face’s expansion into physical AI.
📌 Read More: https://techcrunch.com/2025/04/14/hugging-face-buys-a-humanoid-robotics-startup/

Humans Struggle with Deepfake Voices

A new study reveals that people are only slightly better than chance at identifying AI-generated voices. Participants correctly spotted deepfake audio just 60% of the time, underscoring how realistic synthetic voices have become and raising concerns around audio misinformation.
📌 Read More: https://www.newscientist.com/article/2476556-people-are-really-bad-at-spotting-ai-generated-deepfake-voices/

Nvidia to Build AI Supercomputers in the US

Nvidia will manufacture advanced AI chips and supercomputers in the US, partnering with TSMC, Foxconn, and Wistron. It’s part of a $500 billion investment to localize AI hardware and reduce reliance on foreign manufacturing.
📌 Read More: https://blogs.nvidia.com/blog/nvidia-manufacture-american-made-ai-supercomputers-us/

Meta Will Train AI on EU User Data

Meta plans to train its AI using publicly shared content from Facebook and Instagram users in the EU. Users over 18 can opt out. Meta says the data will help fine-tune AI models to better understand European languages and culture.
📌 Read More: https://www.theverge.com/news/648128/meta-training-ai-eu-user-data

 AI Tools to check out

1. Ask Steve: 1-Click AI Assistent and Page Scraper

2. Backsy: Turn raw feedback into actionable insights

3. AI Barbie: Create Personalized AI Barbie Dolls with AI Barbie

4. SEOJuice: Finds all the contextual internal linking opportunities & more

Thanks for sticking with us to the end!

We'd love to hear your thoughts on today's email!

Your feedback helps us improve our content

⭐⭐⭐Superb
⭐⭐Not bad
⭐ Could've been better

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Tom