Arcee AI

Arcee AI: Small Language Models Enterprises Can Own

Arcee AI is a US-based open intelligence lab that builds small language models (SLMs) and the platform to train, customize, and deploy them. Rather than renting a trillion-parameter frontier model, Arcee's pitch is compact open-weight models — its Trinity family spans Trinity Large Thinking for advanced reasoning, Trinity Mini for balanced workloads, and Trinity Nano for low-latency and edge use — that organizations can download, fine-tune, and run inside their own infrastructure.

It is aimed at enterprises in regulated industries (legal, healthcare, insurance, finance) and at teams building agents that need controllable, affordable models. You can call Trinity models through the Arcee platform API, self-host the open-weight releases, try them free at chat.arcee.ai, or use Arcee's training tooling — including its widely used model-merging work — to produce a domain-specific model of your own.

Key Features

• Trinity model family: Large Thinking (reasoning), Mini (balanced), and Nano (lightweight/edge), released open-weight
• Arcee Platform with API access and self-hosting options for building agents on Arcee models
• Open-source catalog and training tooling, including MergeKit for model merging and Spectrum for efficient training
• Domain-specific SLM training for enterprises that want models tuned on their own data in a secure environment
• Free chat interface at chat.arcee.ai for evaluating the models
• Deployment flexibility from cloud API to fully in-VPC or edge installs

Ideal Use Case

An enterprise team in a regulated industry that needs GenAI capability but cannot ship sensitive data to a closed frontier API: pick a Trinity model, fine-tune or merge it on internal data, and deploy it in-VPC — getting most of the task performance at a fraction of the inference cost, with full ownership of the weights.

How Arcee AI differentiates

Arcee has bet on small, open, ownable models since before it was fashionable and has the backing to sustain it: a $24M Series A led by Emergence Capital closed just six months after its $5.5M seed, followed by a later strategic round to accelerate its AFM foundation-model platform. Its open-source MergeKit became a standard tool for model merging across the community, and press coverage from TechCrunch and VentureBeat has positioned its Trinity releases as rare US-made open-weight models enterprises can download and customize.

FAQ

What is Arcee AI? Arcee AI is a lab and platform for small language models — it trains open-weight models like the Trinity family and helps enterprises deploy or customize them.

Are Arcee's models open? Yes, Trinity models ship open-weight with an open-source catalog, alongside the hosted platform API.

How can I try Arcee models? A free chat interface is available at chat.arcee.ai; developers can use the platform API or self-host the weights.

Who is Arcee AI for? Enterprises in regulated industries and teams that want domain-tuned, self-hostable models instead of closed frontier APIs.

tl;dr

Arcee AI builds open-weight small language models (the Trinity family) plus the training and deployment platform around them — backed by a $24M Emergence-led Series A and the team behind MergeKit.

Related

Looking for more options? Browse the AI/ML Models directory or read our best AI models listicle. Arcee AI is also tracked on Crunchbase.