May 23, 2025

This week, we're excited to roll out significant enhancements to Memex, bringing you more power and flexibility. We're introducing support for two state-of-the-art LLMs, Claude Sonnet 4 and Gemini 2.5 Pro, and laying the groundwork for a more interconnected AI ecosystem with Model Context Protocol (MCP) support. Try them out today!

TLDR

• Claude Sonnet 4: Now the core model powering Memex, offering best-in-class performance for most tasks

• Gemini 2.5 Pro: Available in Beta Long Context Mode with 1M context window, excellent for complex debugging and long conversations

• MCP Templates: Three ready-to-deploy templates for building MCP servers

• MCP Client Integration: Working toward making Memex a full MCP client

1. Bringing the Best Models to Memex

At Memex, we've always been thoughtful about the models we integrate. Our philosophy, as David recently shared, is not to chase quantity but quality. We focus on the 2-3 best SOTA models at any given time because we know that's what delivers the best results for coding and complex tasks. Today, we're thrilled to expand our model offerings in line with this principle, and enable you to try these two brand new models from a single interface!

Claude Sonnet 4: Our New Core Model

Claude Sonnet 4 is now the default model powering Memex. Released this week by Anthropic, it represents a significant leap forward in coding capabilities and instruction following.

It particularly shines in areas like file editing, where its precision is remarkable, and in design-related tasks, offering aesthetically pleasing and functional UI suggestions. If you're working on front-end UI or need nuanced text generation, Claude Sonnet 4 is your go-to. Sonnet 4 also excelts at maintaining focused effort across complex workflows.

Gemini 2.5 Pro: Unlocking Long Context

Gemini 2.5 Pro is now available as part of our Beta Long Context Mode, offering a 1M context window with plans to expand to 2M tokens. We've found Gemini particularly strong for system-level tasks, debugging, and quantitative work.

The expanded context window addresses a fundamental challenge with AI coding assistants: maintaining project context across long conversations. When you hit context limits with Claude, you can seamlessly switch to Gemini and continue working.

Our internal testing shows Gemini 2.5 Pro excels at:

• Complex debugging scenarios

• System-level programming tasks

• Quantitative analysis and computational logic

• Long-running conversations that require maintaining context

If you find yourself stuck on a stubborn bug with another model, switching to Gemini 2.5 Pro significantly increases the chances of a swift resolution. While Claude excels at file editing, Gemini's strength lies in generating sophisticated implementations, even if it sometimes requires a quick syntax check.

2. Introducing MCP Templates

The Model Context Protocol represents a significant step toward more extensible AI systems. We're excited to begin supporting MCP with three templates and a roadmap toward full client integration.

Available Templates

1. Build and Deploy an MCP Server with Netlify: This template provides a minimal, serverless MCP server that can be easily deployed to Netlify. It includes a basic "run analysis report" tool and a resource with documentation on interpreting reports. It's a great starting point for understanding MCP fundamentals and deploying simple tools quickly.

2. Cloudflare MCP Boilerplate: For more advanced use cases, this template offers a robust MCP server boilerplate designed for Cloudflare. It comes with built-in user authentication (Google/GitHub) and Stripe integration, allowing you to create and monetize your MCP tools with one-time, subscription, or metered billing.

3. Streamable HTTP MCP Server: This template provides a FastMCP 2.0 implementation using the Streamable HTTP protocol, designed for deployment on Render. It's a good fit if you need to build more complex MCP servers that require persistent data storage, and the use of Streamable HTTP makes it suitable for tools that might benefit connecting to cloud APIs.

More MCP Features Are in the Oven

These templates are just the beginning. We're working on making Memex a full MCP client. This means you'll soon be able to:

• Connect to MCP servers easily, without any complicated setup.

• Tap into a growing world of MCP tools and resources.

• Build and share your own MCP integrations smoothly.

This approach will make Memex more powerful and helpful, opening up connections to a wider array of services and tools.

What This Means for You

These updates are all about bringing you a more powerful Memex:

• Better everyday performance: With Claude Sonnet 4 as our core model, you'll see improvements in coding and how well Memex follows instructions for your daily tasks.

• Tackle bigger projects: Gemini 2.5 Pro's long context window means you can now work on larger, more complex challenges.

• Build your own extensions: Our new MCP templates give you a starting point for creating custom tools and integrations.

By combining top-tier models with open protocols, we're moving closer to AI that can truly boost what you can do, no matter what you're working on.

As always, we welcome your feedback on these updates. Join our Discord community or reach out at help@memex.tech.

On a more informal note…

…our team got together in London earlier this month to plot out the next couple of months for Memex. Thanks for being part of the journey!