# Arcee AI

> Arcee AI is a San Francisco-based open intelligence lab that builds small, efficient, US-made language models enterprises can own, customize and run inside their own infrastructure. Founded in 2023 by former Hugging Face and Roboflow operators, the company pioneered practical techniques like model merging (via its acquisition of mergekit) and Spectrum training, and has expanded from small language models into its own open-weight Arcee Foundation Model family, the Trinity mixture-of-experts models, and agentic products such as Arcee Conductor and Arcee Orchestra. Its pitch: smaller, domain-adapted models that rival frontier systems at a fraction of the cost while keeping data private.

- **Founded:** 2023
- **Headquarters:** San Francisco, United States
- **Founders:** Mark McQuade (Co-founder & CEO), Jacob Solawetz (Co-founder & CTO), Brian Benedict (Co-founder & CRO)
- **Team size:** ~39 employees (2024-2025); reported to have operated with a very small core team (~26) while training large models
- **Products:** MergeKit, Spectrum, Arcee Enterprise, Arcee Cloud, SuperNova
- **Notable:** Raised $24M Series A just six months after a $5.5M seed, signaling investor confidence in small language models., Acquired the popular mergekit open-source toolkit and hired its creator., Shipped its own foundation model family (AFM) and the open-weight Trinity mixture-of-experts models, released under Apache 2.0.

## Products & services

- **MergeKit** — Open-source toolkit for merging multiple large language models into a single more capable model without additional training; acquired by Arcee and led by its creator Charles Goddard.
- **Spectrum** — A training technique that targets the most informative layers of a model to cut training time and compute while preserving quality.
- **Arcee Enterprise** — In-VPC deployment of Arcee's platform for training, merging, deploying and monitoring small language models inside a customer's own cloud.
- **Arcee Cloud** — Hosted SaaS version of the Arcee platform that brings SLM training and deployment to a broader set of teams.
- **SuperNova** — A 70B-parameter model designed for enterprise deployment, emphasizing ownership, data privacy, stability and customizability.
- **AFM-4.5B (Arcee Foundation Model)** — The first Arcee Foundation Model, a 4.5B-parameter model marking Arcee's shift to building its own foundation models.
- **Arcee Conductor** — SaaS platform that routes queries to the most appropriate small model, where Arcee's production models were battle-tested before open release.
- **Arcee Orchestra** — End-to-end agentic AI product for building custom workflows and AI agents that route tasks to specialized small language models.
- **Trinity model family** — Open-weight attention-first mixture-of-experts (AFMoE) models trained in the US - including Trinity Nano (~6B), Trinity Mini (~26B) and Trinity Large (~420B/13B active), released under Apache 2.0.

## Achievements

- Raised $24M Series A just six months after a $5.5M seed, signaling investor confidence in small language models.
- Acquired the popular mergekit open-source toolkit and hired its creator.
- Shipped its own foundation model family (AFM) and the open-weight Trinity mixture-of-experts models, released under Apache 2.0.
- Built a 400B+ parameter open model with a very small team, positioning as a US-made open-source alternative to Big Tech models.
- Hosts 200+ open models on Hugging Face with a large developer following.

## Latest updates

- **2025-12** — Released the open-weight Trinity model family (Trinity Mini ~26B and Trinity Nano preview ~6B) under Apache 2.0, with Trinity Large (~420B/13B active) in training and slated for early 2026.
- **2025-07** — Deep-dived AFM-4.5B, the first Arcee Foundation Model, marking the shift to building its own foundation models.
- **2025** — Expanded from SLM tooling into agentic products - Arcee Conductor (model routing) and Arcee Orchestra (agent workflows).
- **2024-07** — Closed $24M Series A led by Emergence Capital with Notable Capital, and launched Arcee Cloud.

## Links

- Website: https://www.arcee.ai
- LinkedIn: https://www.linkedin.com/company/arcee-ai
- Twitter/X: https://x.com/arcee_ai
- GitHub: https://github.com/arcee-ai
- YouTube: https://www.youtube.com/@ArceeAI

---

Profile page: https://yespress.io/arcee-ai
Published by YesPress — https://yespress.io
Last updated: 2026-07-13