cto.new

Executive Summary

cto.new, built by Engine Labs, is an AI coding agent that provides access to frontier language models without requiring a credit card or API key. Launched in October 2025 as "the first completely free AI code agent," the platform gained early traction with developers at companies including Adobe, Docker, Twilio, and PwC. As of June 2026 the model has shifted to freemium: the free tier is ad-supported with limits that reset daily, and a Premium plan starts at $20/month. The product has also expanded beyond background coding into multi-agent teams, an AI SRE agent, and a marketplace for ready-made AI teams.

---

Product Overview

cto.new launched as the "completely free AI code agent" — a bold claim in a market where competitors like Claude Code and Cursor charge $20-200/month for comparable capabilities. The platform handles the full development workflow: planning, code generation, testing, and creating pull requests. By mid-2026 the positioning broadened from coding agent to "build AI applications, agents, and startups," with multi-agent teams and an AI team marketplace.

The product separates work into two modes: Chat for planning and ideation, and Tasks for actual code execution. This separation mirrors how senior developers work — thinking through problems before implementing solutions.

Key Capabilities

Product Surfaces

Launches Since February 2026

• AI SRE agent (announced on the company blog, February 26, 2026)
• Backend on Convex — "Why we bet on Convex" engineering post (March 3, 2026)
• Free CLI and multi-agent teams / marketplace, now featured on the homepage (as of June 2026)
• Premium tier starting at $20/month, replacing unlimited free usage with daily-reset limits on the free plan

---

Technical Architecture

cto.new runs code agents in sandboxed environments, recently migrating from self-hosted Firecracker microVMs to Daytona's managed infrastructure. The platform prioritizes pseudo-terminal support for realistic LLM interactions.

Key Technical Details

Agent Toolset

The cto bench benchmark reveals the platform's core tools:

• ReadFile — Reads text files from absolute paths
• WriteFile — Writes/overwrites files
• EditFile — Replaces text blocks in existing files
• GlobTool — Finds files by glob patterns
• GrepTool — Searches file contents with regex
• LsTool — Lists directories with pagination
• TerminalTool — Runs shell commands in a VM terminal

---

Strengths

• Low Cost Barrier — No credit card or API key required to start. Even with daily limits now on the free tier, cto.new remains one of the most accessible ways to try AI coding agents.

• Frontier Model Access — Users get access to the same Anthropic and OpenAI models that competitors charge premium prices for, with routing across providers. The token subsidy strategy may not last forever, but early adopters benefit massively.

• Proven Scale — Processing 300+ billion tokens and shipping to 10,000+ codebases in the first three months demonstrates the platform can handle production workloads, not just demos.

• Real-World Benchmarking — The cto bench approach uses actual task completion rates rather than synthetic tests, providing more meaningful model comparisons for coding tasks.

• IDE Integration via MCP — The ability to connect cto.new as a background agent from Cursor addresses a key pain point: Cursor's AI features can be expensive for heavy users.

---

Cautions

• The "Completely Free" Promise Already Eroded — Within roughly four months of this profile's original date, the unlimited free tier became ad-supported with daily-reset limits plus a paid Premium plan from $20/month. This validates the sustainability skepticism voiced at launch on Hacker News, where one commenter (yodon) wrote: "Offering frontier models for free at a Claude Code Max or ChatGPT Pro levels isn't going to happen for very long with a $5.7M raise."

• Sustainability Questions — The "Making Tokens Affordable" blog post openly acknowledges that "LLM inference is quite expensive" and the free model requires monetization through data collection and ads. The long-term viability of generous free access remains uncertain.

• Privacy Trade-offs — Free access comes at a cost. The company says the product is free "because we use anonymized usage data for model training and other purposes." Enterprise users with proprietary code should evaluate this carefully.

• New Company Risk — Launched October 2025, cto.new has limited operating history. The rapid scope expansion (coding agent → AI businesses, agent teams, marketplace) could signal either ambition or a search for a durable business model.

• Limited Visibility — Founder backgrounds and team composition aren't prominently featured, making due diligence difficult for enterprise evaluations.

• Compliance Claims Unverified — Earlier versions of the pricing page mentioned SOC2 compliance; as of June 2026 the pricing page no longer surfaces compliance details, and no public attestation reports are linked.

---

What Developers Say

From the Hacker News launch thread (October 2025, 11 points, 6 comments) — a small discussion, dominated by skepticism:

"Offering frontier models for free at a Claude Code Max or ChatGPT Pro levels isn't going to happen for very long with a $5.7M raise. It's hard to imagine a scenario where this raise isn't completely burned through in a couple of months, tops." — yodon

"free AI coding agents but 'completely for free because we use anonymized usage data for model training and other purposes.'" — charv

"When something is 'free', YOU are the product!" — bamazizi

Beyond the launch thread, organic community discussion (Hacker News, Reddit) remains thin as of June 2026 — coverage is mostly press releases and AI tool directories. The company's own adoption claim of "more than 6,000 developers at Adobe, Docker, Twilio and PwC" comes from its funding announcement and has not been independently verified.

---

Pricing & Licensing

Pricing as of June 2026, from cto.new/pricing.

Licensing model: Freemium (free tier is ad/data-supported; Premium subscription lifts daily limits)

Hidden costs: Your code and usage data contribute to cto.new's monetization strategy on the free tier.

---

Competitive Positioning

Direct Competitors

When to Choose cto.new Over Alternatives

• Choose cto.new when: You want to experiment with AI coding agents without budget commitment, or you're a solo developer comfortable with data-for-service trade-offs
• Choose Claude Code when: You need deep integration with Anthropic's ecosystem and have budget for subscriptions
• Choose Cursor when: You prefer local IDE experience and want more control over your development environment
• Choose Tembo when: You need orchestration across multiple coding agents at enterprise scale

---

Ideal Customer Profile

Best fit:

• Solo developers and indie hackers evaluating AI coding tools
• Students and learners wanting to experience frontier AI capabilities
• Developers who want to offload routine coding tasks without subscription costs
• Teams using Cursor who want to reduce AI feature costs via MCP integration

Poor fit:

• Enterprises with strict data governance requirements
• Companies with proprietary code they can't risk exposing to third parties
• Organizations requiring verifiable compliance certifications
• Users who need guaranteed service availability and SLAs

---

Viability Assessment

The $5.7M seed round suggested investor confidence, but the aggressive free pricing created a high burn rate — and the predicted pricing-model change arrived within months. The free tier now carries daily limits and ads, with Premium from $20/month. Watch for a Series A announcement or further tightening of free limits as leading indicators.

---

Bottom Line

cto.new launched as an aggressive bet that AI coding agents could be ad/data-supported rather than subscription-based — and within months pivoted to freemium, exactly as skeptics predicted. The free tier (ad-supported, daily limits) is still one of the lowest-friction ways to try a background coding agent, and the Premium plan at $20/month is priced in line with Cursor and Claude Code's entry tiers rather than below them.

Recommended for: Individual developers, students, and indie hackers who want to experiment with AI coding agents without budget commitment and are comfortable with data-for-service trade-offs.

Not recommended for: Enterprises with proprietary code, organizations requiring compliance verification, or users who need guaranteed service availability.

Outlook (as of June 2026): The freemium pivot resolved the most acute sustainability question but erased the headline differentiator. cto.new's future now depends on whether its broader platform play — multi-agent teams, an AI SRE agent, and an AI team marketplace — can carve out an identity beyond "the free one." The company is shipping fast; the open question is focus.

---

Research by Ry Walker Research • methodology