Signup  |  Work With Us  |  Follow on X  |  Read on Web

.

Hey ,

Welcome to AlphaSignal – the most read newsletter by AI developers. 


We bring you the top 1% of news, papers, models, and repos, all summarized to keep you updated on the latest in AI.

IN TODAY'S SIGNAL

Read time: 6 min 02 sec

🎖️ Top News

📌  Aiola

⚡️ Trending Signals

💻  Top Repos




  • Agno helps build multimodal AI agents with memory, tools, and reasoning, running 10,000x faster than LangGraph.


  • GitDiagram: Convert any GitHub repository into an interactive system design diagram with OpenAI-powered insights and navigation.


  • MCP-Agent is a lightweight framework for building AI agents using the Model Context Protocol (MCP).


🧠  Top Lecture

If you're enjoying AlphaSignal please forward this email to a colleague. 

It helps us keep this content free.

TOP NEWS

AI Model

Google releases Gemini 2.5 Pro for all with Canvas mode, enhancing real-time coding and app development


⇧ 7,438 Likes

What's New

Google now provides free access to Gemini 2.5 Pro (Experimental) through the Gemini app and AI Studio. The model ranks as the top AI system on LMArena and Devin internal evaluations. You can now use it without a subscription, requiring only a Google account.


New Developer Features
Gemini 2.5 Pro introduces updates that improve usability in AI Studio.

  • Canvas mode enables interactive, structured coding and iterative development.

  • Shows real-time updates and responses inside a live coding workspace

  • Supports prompt-to-app workflows with editable UI

  • “Get code” includes full-screen view and toggleable prompt history

  • Use in AI Studio’s live mode with output in executable code format

  • "Open in Colab" retains prompt settings for seamless transitions.

  • History toggle and fullscreen mode improve workflow visibility and navigation.


Benchmark Performance Updates
The model ranks at the top of multiple AI benchmarks after the release.

  • #1 on LMArena for overall AI performance

  • #1 on Devin internal evaluations for reasoning and coding tasks

  • 130 IQ on Mensa Norway in problem-solving and analytical assessments

  • Outperforms Claude 3 and GPT-4 in STEM single-task evaluations

  • Falls behind GPT-4 in multi-turn chat and coding task accuracy

API Access and Rate Limits
You can integrate Gemini 2.5 Pro through API access with some restrictions.

  • 20 requests per minute (RPM) with billing enabled.

  • Free access limited to AI Studio and in-app usage.

  • No announced fine-tuning or model customization options.

  • Performance optimized for reasoning, multimodal tasks, and coding.



Supported Platforms
Gemini 2.5 Pro is accessible across multiple platforms.

  • Available in Cursor and Windsurf for coding-related tasks.

  • Integrated into Google AI Studio with Canvas mode support.


Community Feedback

Chris Wall
"I have been using Gemini 2.5 Pro alll weekend weeekdn and I am blown away at how good it is at coding. It is truly astonishing what I have been able to build with it in a weekend. Great work"

Andy Wojcicki
"For building something we need higher rate limits, or it will remain only a toy. Especially for the new models (image gen, 2.5) it won't be possible to release anything to a broader audience."

Ben Pielstick
"I guess Google has more GPUs for this than OpenAI."

TRY NOW

Need an Accurate Enterprise ASR built for Real-World Use?

aiOla’s new foundation speech-to-text model, Jargonic, delivers enterprise-grade accuracy in real-world conditions.

It beats OpenAI, ElevenLabs, Deepgram, and AssemblyAI with a 5.91% average Word Error Rate and over 97% jargon recall in English, Spanish, Portuguese, German, and French.


Features include:

  • 1M+ hours of training data: Handles noisy environments, domain-specific jargon, and accents with zero-shot keyword spotting.

  • APIs for speech processing – Speech-to-text, text-to-speech, and workflow automation.

  • Real-time data structuring – Converts spoken data into structured reports and notifications.

Capture and process spoken data with unmatched accuracy.

GET EARLY ACCESS

partner with us 

TRENDING SIGNALS

Web Automation

Amazon introduces Nova Act, an AI model trained to perform actions within a web browser, beats Claude 3.7 in web UI benchmarks

⇧ 1,036 Likes

AI Business

OpenAI secures historic $40B round led by SoftBank to push AGI development and AI tools

⇧ 7.946 Likes

Generative AI

Runway announces Gen-4, a 1080p video generation model with controlled character, object, and environment consistency

⇧ 3,257 Likes

AI Tool

Ai2 unveils CodeScientist, an AI system for autonomous scientific discovery through literature analysis and experiment execution

⇧ 493 Likes

Speech AI

ElevenLabs unveils Actor Mode in their AI Studio, offering precise control over AI-generated voice with custom pacing and emphasis

⇧ 1,738 Like

TRENDING REPOS

Agents

agno

☆ 23,447 Stars

It helps you build multimodal AI agents that generate text, images, audio, and video. It runs 10,000x faster than LangGraph with 50x lower memory use. Agents use memory, tools, and reasoning. Support structured outputs, RAG via vector DBs, real-time monitoring, and multi-agent coordination. Works with any model or provider.

AI tool

gitdiagram

☆  6,037 Stars

Use GitDiagram to generate interactive architecture diagrams from any GitHub repository. Replace “hub” with “diagram” in the URL. Uses OpenAI o3-mini for structured insights. Supports customization, private repos, and self-hosting. Built with Next.js, FastAPI, and PostgreSQL. Deployed on Vercel and EC2. No rate limits currently.

Agents

mcp-agent

☆ 2491 Stars

Use this repo to build AI agents using the Model Context Protocol with composable workflow patterns. Connect to MCP servers, manage tool lifecycles, and compose patterns like Router, Orchestrator, Swarm, and Evaluator-Optimizer. Each workflow wraps into AugmentedLLM with memory, tool calls, and structured outputs. Supports Claude, OpenAI, and more.

TOP LECTURE

MCP

Using MCP Servers with Google Gemini 2.0

⇧ 967 Likes

This guide explains how to integrate Gemini models with MCP servers for structured AI tool interactions. MCP standardizes access to external tools and data sources, making it easier to build AI applications without custom integrations.

You will learn:

  • How MCP provides structured tool access via JSON schemas

  • How to convert MCP tools to Gemini-compatible OpenAPI schemas

  • Implementing tool calling in Gemini models using MCP

  • Running an agent loop for automated tool execution

  • Handling function calls and responses within Gemini

  • Using MCP for real-world AI workflows, such as booking services


This tutorial includes Python code examples to help you implement MCP-based integrations efficiently.

TRY NOW

Stop receiving emails here