Kilo CLI 1.0 brings vibe open source code to your terminal with support for 500+ models

Remote AI coding startup Kilo doesn’t think software developers have to swear their undying loyalty to any development environment — and certainly not a single model or harness.

This week, the startup — backed by GitLab founder Sid Sijbrandij — unveiled Kilo CLI 1.0, a complete redesign of its command-line tool that offers support for more than 500 underlying AI models from proprietary leaders and open source competitors like Alibaba’s Qwen.

It comes a few weeks after Kilo launched Slackbot which allows developers to send code directly from Salesforce’s popular messaging service (Slack, also used by VentureBeat) powered by Chinese AI startup MiniMax.

The release marks a strategic pivot away from the “sidebar” IDE model favored by industry giants such as Cursor and GitHub Copilot, or dedicated applications such as the new OpenAI Codex, and even endpoint-based competitors such as Codex CLI and Claude Code, which aim to embed the power of AI into every piece of professional software workflow.

By introducing a model-agnostic CLI on the heels of its Slack bot, Kilo is making a calculated bet: the future of AI development is not about a single interface, but about tools that accompany developers between IDEs, terminals, remote servers, and group chat threads.

In a recent interview with VentureBeat, Kilo CEO and co-founder Scott Breitenother explained the need for this fluid shift: “The experience feels very fragmented right now… as a developer, sometimes I’ll be using the CLI, sometimes I’ll be in VS Code, and sometimes I’ll be kicking an agent, and I have to step out a little bit.”

He noted that Kilo CLI 1.0 is specifically “built for this world… for the developer who is moving between their local IDE, a remote server via SSH, and the last session at 2 a.m. to debug a production bug.”

Technology: Rebuilding ‘Kilo Speed’

Kilo CLI 1.0 is a fundamental architectural change. While 2025 was the year when major developers started to take AI vibe coding seriously, Kilo believes that the year 2026 will be defined by the adoption of agents that can handle end-to-end tasks independently.

The new CLI is built on an MIT-licensed, open-source foundation, designed to work in the last moments when developers often find themselves in the middle of critical production incidents or intensive infrastructure work.

For Breitenother, building in the open source is non-negotiable: “When you build in the open source, you build the best products. You get this great wheel of collaborators… your community is not just passive users. It’s actually part of your team that helps you improve your product… Honestly, some people might say that open source is a weakness, but I think it’s our greatest strength.”

At the heart of this “functionality” is Kilo’s ability to go beyond auto-completion. The CLI supports multiple modes of operation:

• Code mode: For high-speed production and multi-file refactors.

• Architect Mode: For high-level planning and technical strategies.

• Debug mode: For systematic problem analysis and solution.

Solving multi-session memory

To solve the persistent problem of “AI amnesia”—when an agent loses context between sessions—Kilo uses a “Memory Bank” feature.

This system maintains state by keeping context in structured Markdown files within the repository, ensuring that an agent running in the CLI has the same understanding of the codebase as one running in a VS Code sidebar or Slack thread.

The synergy between the new CLI and the “Kilo of Slack” is the basis of the company’s “Agenttic Anywhere” strategy. Launched in January, the Slack integration allows teams to fix bugs and push pull requests directly from the chat.

Unlike competing integrations from Cursor or Claude’s Code – which Kilo claims are limited by the configuration of a single repo or the lack of continuous thread state – Kilo’s bot can import content from multiple repositories at once.

“Development teams don’t make decisions in IDE sidebars. They make them in Slack,” Breitenother emphasized.

The extensibility and ‘superpower’ of open source

An important part of Kilo’s technical depth is its support for the Model Context Protocol (MCP). This open standard allows Kilo to communicate with external servers, extending its capabilities beyond local file manipulation.

With MCP, Kilo agents can integrate with custom tools and resources, such as internal document servers or third-party monitoring tools, turning the agent into a special member of the engineering team.

This expansion is part of Kilo’s commitment to atheism. Although MiniMax is the default for Slack, the CLI and extension support a large array of over 500 models, including Anthropic, OpenAI, and Google Gemini.

Price: The ‘AI output per dollar’ economy

Kilo is also trying to disrupt the economics of AI development with “Kilo Pass,” a subscription service designed for transparency.

The company charges API provider direct rates with zero commission—a $1 Kilo credit equals $1 in provider fees.

Breitenother criticizes the “black box” subscription models used by others in the space: “We’re selling infrastructure here… you hit some kind of illogical, vague line, and you start to shrink. That’s not the way the world is going to work.”

Kilo Pass categories offer “instant rewards,” which provide bonus credits to active subscribers:

• Starter ($19/mo): Up to $26.60 in credits.

• Pro ($49/mo): Up to $68.60 in credits.

• Professional ($199/mo): Up to $278.60 in credits.

To encourage early adoption, Kilo is currently offering a “Double Welcome Bonus” until February 6, giving users 50% free credits for their first two months.

For power users like Sylvain, this flexibility is a big draw: “Kilo Pass is exactly what I’ve been waiting for. I can use my credits when I need them and save them when I don’t—it’s finally compatible with how I use AI.”

Society, security, and competition

The arrival of Kilo CLI 1.0 puts it in direct conversation with terminal-native heavyweights: Anthropic’s Claude Code and Block’s Goose.

Aside from the terminal, in the more crowded IDE space, OpenAI recently launched a new Codex desktop application for macOS.

Code Claude offers a more polished experience, but comes with vendor lock-in and high fees—up to $200 per month for titles that still include token-based usage caps and rating limits. Independent analysis suggests that these limits are often exhausted within minutes of hard work on large codes.

The new OpenAI Codex program similarly favors a platform-locking approach, acting as an “agent command center” that allows developers to direct autonomous AI systems for up to 30 minutes.

While Codex introduces powerful features like “Skills” to connect to tools like Figma and Linear, it’s basically designed to protect the OpenAI ecosystem in a highly contested market.

In contrast, Kilo CLI 1.0 uses the MIT-licensed OpenCode foundation to deliver a production-ready Terminal User Interface (TUI) that allows developers to switch between 500+ models.

This portability allows teams to choose the best cost-to-performance ratio—perhaps using a lightweight model for documentation but switching to a boundary model to remove complex error.

In terms of security, Kilo ensures that models are hosted on US-compliant infrastructure such as AWS Bedrock, allowing proprietary code to stay within trusted perimeters while using the most efficient intelligence available.

Goose provides an open source alternative that runs entirely on the user’s local machine for free, but it appears to be localized and experimental.

Kilo positions itself as a middle ground: a robust productivity tool that maintains open source visibility while providing the infrastructure to reach the entire enterprise.

This contrasts with two broader market concerns; while OpenAI builds sandboxes to protect autonomous agents, Kilo’s open environment allows for a “high-powered” level of public experimentation and contribution.

The future: The ‘mech suit’ of the mind

With $8 million in seed funding and a “Right of First Refusal” agreement with GitLab that lasts until August 2026, Kilo is positioning itself as the backbone of the next-generation developer stack.

Breitenother views these devices as “exoskeletons” or “mech suits” for the mind, rather than replacements for human engineers.

“We’ve actually moved our engineers into product ownership,” Breitenother revealed. “When they’re comfortable writing code, they’re actually thinking a lot. They’re setting a product strategy.”

By deconstructing the engineering stack—separating the agent interface from the model and the model from the IDE—Kilo provides a roadmap to a future where engineers think about architecture while machines build architecture.

“It’s the closest thing to magic I think we’ll ever encounter in our lives,” Breitenother concluded. For those looking for “Kilo Speed,” the IDE sidebar is the starting point.