article

AI Dev Essentials #10: Cursor Hits 1.0 (and $9.9B!), OpenAI o3/pro, Gemini Updates

AI Dev Essentials #10: Cursor 1.0 lands with $900 M, OpenAI cuts o3 pricing & debuts o3-pro, Gemini 2.5 Pro tops coding tests—plus fresh tools, funding updates

Hey Everyone 👋,

John Lindquist here with the tenth issue of AI Dev Essentials! This week has been absolutely monumental for the AI coding space. Cursor hitting 1.0 and raising $900 million at a nearly $10 billion valuation is a watershed moment for AI-assisted development. The speed at which they've grown—from $1 million to over $500 million in ARR in just over two years—shows just how hungry developers are for better AI coding tools. And with background agents now generally available, we're entering a new era where AI can truly work alongside us asynchronously.

I've personally spent a lot of time withe the "Claude Code" experience the past week to compare it to Cursor. I totally understand the appeal, especially because of how reliably Claude Code can churn away at long tasks (and you're not thinking about burning through requests/costs), but Cursor is still a superior experience when it comes to gathering context and reviewing changes. I'm hoping this puts pressure on the Cursor team to release a "MAX" plan similar to Claude's "MAX" plan. I like having a budget, but I don't like thinking every failed tool call is lost money. I try to remember that we're still all in the super early days of AI, and everyone is just figuring it out as they go.

With that said, let's dive into this week's updates!

🎓 New egghead.io Lessons This Week

I'm working on a course "Building Local AI Agents with Ollama and the AI SDK". I'm 5 lessons in and I'm excited to share them, but just not quite ready. They should be up when this newsletter rolls around next week, so keep an eye out for that. 😇

🚀 Cursor 1.0 & The $900M Funding Round

Cursor has officially hit 1.0 and raised $900 million at a $9.9 billion valuation

This week brought two massive announcements from Cursor that cement its position as the leader in AI coding assistants. First, the 1.0 release brings game-changing features like BugBot for automated code review, background agents for asynchronous work, and seamless GitHub integration. Then came the funding news: $900 million led by Thrive Capital, with participation from Andreessen Horowitz, Accel, and DST Global.

Key Stats:

Cursor 1.0 Features:

(TechCrunch Coverage, DEVCLASS Analysis)

I do love seeing money being poured into companies that are building AI tools for developers. I'm hoping that the leads to customized models that are smarter, faster, with reliable tool calling, and features around "agent swarms" that can tackle complex tasks in parallel. I'm really excited to see what sorts of UIs emerge for orchestrating these agents.

🧠 Model & Platform Updates

OpenAI o3 & o3-pro: Major Price Drop and Release

OpenAI made significant waves this week with two major announcements. First, Sam Altman confirmed an 80% price reduction for o3, making it viable as a "daily driver" model. Cursor immediately updated their pricing to reflect this change, noting that "o3 is a fantastic model, and we're excited that it can now be used as a daily driver."

Then came o3-pro, OpenAI's most advanced reasoning model yet. Community consensus is forming around its capabilities and limitations:

o3-pro Strengths:

Community Reality Check:

Codex Gets Internet Access: OpenAI also rolled out internet access for Codex, allowing it to install dependencies, run tests with external resources, and upgrade packages during task execution. The feature is off by default and provides granular control over allowed domains and HTTP methods.

This all rolled out last night. I haven't had a chance to dig in to o3-pro, so I just gathered a bunch of early impressions from the community above. I'll have my own personal take next week.

Gemini 2.5 Pro Tops WebDevArena

Google's updated Gemini 2.5 Pro continues to excel at coding, now leading WebDevArena with a 1443 Elo score (35-point jump). The update introduces:

(Google Blog, 9to5Google)

Gemini 2.5 Pro is in this interesting place where it's awesome, but still a bit difficult to work with in Cursor. I really, really want to use it as my daily driver, but the thinking process often seems to spin too much. I'm hoping a Cursor system prompt (or a community propmt) will emerge to help smooth it all out. The results have always been great, but sometimes it ends up taking too long and my patience wears out.

Mistral Launches Magistral: Multilingual Reasoning Models

Mistral AI announced Magistral, their first reasoning models designed to excel in domain-specific, transparent, and multilingual reasoning. The models come in two variants:

Key differentiators include multilingual reasoning (especially strong in European languages), 10x speed improvements in Le Chat, and transparent thinking processes users can follow and verify. (CNBC Report, TechCrunch)

Any open-source news is good news. While I'm happy to pay providers for "premium" access to models, I'm still hopeful open-source will unlock AI for everyone down the road. I just never make time to test the open-source models with so much else happening at the same time.

💰 Major Funding & Business Updates

Linear Raises $82M at $1.25B Valuation for AI Agent Platform

Linear secured $82 million in Series C funding led by Accel, positioning itself as "the place where teams, AI, and agents build software together." With 280% profit growth and over 15,000 customers including OpenAI, Scale AI, and Perplexity, Linear is focusing on making their platform the natural home for AI agents to coordinate work across product workflows. (Reuters, Linear Blog)

I included this because Linear's MCP is a popular MCP for agents. So far, everyone uses so many variations of "AI Task Tracking" and it seems like Linear is raising to the top. Seeing them focus on AI-specific language and raising money is only a good sign (maybe Cursor will buy them or spin off their own version someday?)

🛠️ Developer Tools & Ecosystem

Vertex AI Studio Gets Major Redesign

Google Cloud unveiled a redesigned Vertex AI Studio with a developer-first approach:

(Google Cloud Blog)

ElevenLabs Launches v3 with Emotion Control

ElevenLabs v3 introduces revolutionary emotion direction through audio tags:

(ElevenLabs v3, Documentation)

These voices are so good. I'm tempted to toss one of my favorite books in their ("The Silent Gondoliers") and go on a road trip to see how they handle full narration.

OpenAI Agents SDK Now in TypeScript

OpenAI released their Agents SDK in TypeScript with full feature parity to Python:

(OpenAI Developers)

TypeScript is arguably the language for building AI agents thanks to all the hosting/endpoints/etc already built for it. For most platforms, it becomes trivial to plop in a TypeScript SDK package, an API key, and spin up Agents.

🔧 Quick Updates

🤖 AI Research & Observations

AIs Play Diplomacy: Revealing Model Personalities

A fascinating experiment had different AI models play the strategy game Diplomacy, revealing distinct behavioral patterns:

This real-world test highlights how different models exhibit varying levels of strategic thinking, deception capability, and goal-oriented behavior—important considerations when choosing models for different applications.

I'm hoping someday these can turn into live visual competitive storytelling experiences. Kind of like a battlebots tournament, but with AI models. I'll admit I'm also hiding my head in the sand when it comes to using AI to "dominate" anything. (I love you AI Overlords! ❤️)

Anthropic Cuts Windsurf API Access

In a surprising move, Anthropic reportedly restricted Windsurf's API access to Claude models. This development raises questions about the competitive dynamics between AI providers and coding assistant platforms. (TechCrunch)

Seems fair IMO. OpenAI bought Windsurf. OpenAI competes with Anthropic. Why would Anthropic want to help OpenAI make money?

💡 Security & Best Practices

Vercel on AI Agent Security: New guidance on prompt injection vulnerabilities emphasizes: assume compromise, limit tool call access, and don't trust output. Essential reading for anyone building AI agents. (Vercel Blog)

✨ Workshop Spotlight: Conquer the Complexity of Cursor ✨

Master practical AI development workflows in Cursor. This hands-on workshop covers Agents, Ask mode, Custom Modes, multi-file analysis, effective prompting, Cursor rules, and strategies for handling AI failures.

➡️ Read More about the Workshop

➡️ Register Now ($249)

(Team training also available)

That's all for this milestone tenth issue! The convergence of massive funding, breakthrough features, and growing enterprise adoption shows we're at an inflection point for AI-assisted development. The tools are getting more powerful, the agents more capable, and the possibilities more exciting.

If you have any feedback or questions, hit reply. I'm happy to chat about the latest in AI dev tools.

John Lindquist

egghead.io