Arcee — a 26-person startup — created a powerful open source LLM and is gaining users

Arcee — an American startup with a team of just 26 people — has created a high-performance open source language model that competes in capabilities with products from significantly larger organizations. In an industry where it is customary to believe that powerful LLMs are accessible only to companies with billion-dollar budgets and hundreds of engineers, this story sounds almost unbelievable. And yet — the model works and is already gaining a real audience among OpenClaw users.

The company Arcee was founded as a specialized laboratory for open language models. Its strategy is the direct opposite of what OpenAI, Anthropic, and Google do: instead of closed systems accessible only through paid APIs, Arcee releases models with fully open source code. Developers and companies can download them, adapt them for specific tasks, and deploy them on their own servers — without dependence on external providers, without monthly subscriptions, and without their users' data leaving the corporate infrastructure.

This approach is especially important for the corporate B2B segment. Large companies in finance, law, and healthcare are increasingly demanding control over where their data is processed. Regulatory restrictions in many industries make it practically impossible to send sensitive corporate data to third-party servers.

Closed API providers cannot always meet these requirements — open source models with the ability to deploy locally can. Over the past few years, the methods for creating and adapting language models have been significantly democratized. More efficient fine-tuning techniques, knowledge distillation, and preference alignment have emerged, allowing small teams to achieve results that previously required multi-million-dollar computing budgets.

Open libraries, evaluation tools, and extensive public datasets have become universally accessible infrastructure. It is under these conditions that a team of 26 people was able to create a model that attracts attention. The fact that Arcee is gaining popularity among OpenClaw users is an important practical signal.

This is not academic recognition and not a victory on another benchmark, which tells little about the actual behavior of the model. This is a choice by people who work with language models every day and evaluate them by pragmatic criteria: quality of generation, speed of inference, hardware compatibility, and license cleanliness for commercial application. The story of Arcee fits into a broader trend that has been gaining momentum over the past two years: open source AI is becoming a serious alternative to closed systems.

Meta opened the weights of Llama 3, Mistral is building a business on an open strategy, Alibaba released Qwen, DeepSeek shocked the market with its open releases. Against this backdrop, small specialized teams like Arcee have increasing opportunities: infrastructure is becoming cheaper, expertise is becoming more accessible, and the community around open models is growing. For business, this expands the choice far beyond "pay OpenAI or build from scratch."

Take an open model with a strong reputation, created by a team with real expertise, and adapt it to your task — this is today a real and increasingly common option. The cost of this approach is lower than developing from scratch, and the control over the result is higher than when using a closed API. Arcee is a reminder: in AI, you don't have to be huge to matter.

Sometimes the right choice of openness strategy and a team of 26 people focused on one task turn out to be sufficient to create a product used by thousands.