← All companies

Inference / Company profile

Hugging Face

What Hugging Face does

Hugging Face is a private AI platform centered on the Hugging Face Hub, where developers and organizations publish and collaborate on machine learning artifacts—models, datasets, and interactive applications (Spaces). Its open-source software stack (for example Transformers and the surrounding Hugging Face Hub libraries) helps standardize how models are authored, versioned, evaluated, and shared, while its paid platform and services focus on productionizing those open artifacts for enterprise use. Hugging Face’s offerings extend beyond “hosting”: it provides managed inference deployment (dedicated endpoints), no-code/low-code training tooling (AutoTrain), and packaged/optimized ways to serve open models in customer infrastructure (including HUGS). In strategic positioning, Hugging Face operates as an interoperability layer across major cloud and hardware ecosystems—integrating with infrastructure partners such as Google Cloud and supporting “local AI” trends through investments in inference engines (for example llama.cpp) that complement its cloud and enterprise deployment path.

News

Company record

Aug 14, 2026 · Official · Hugging FaceState of Open Models: Summer 2026 Observations

Hugging Face publishes a new market/ecosystem report, including observations about local inference resources and an agent usage dataset describing agent traffic patterns into the Hub.

Jun 09, 2026 · Official · Hugging FaceArcee partnership: exclusive home for Arcee models, datasets, and agent traces (private storage emphasis)

Hugging Face announces a strategic commercial collaboration with Arcee, stating Hugging Face will be the exclusive home for Arcee’s models, datasets, and agent traces, including private artifacts.

Feb 20, 2026 · Official · Hugging FaceGGML and llama.cpp join Hugging Face to support local AI openness

Hugging Face announces that GGML (creators of llama.cpp) is joining Hugging Face to provide long-term resources for local inference tooling, while retaining open, community-driven governance.

Date not disclosed · Official · Hugging FaceHugging Face launches HUGS (Generative AI Services)

HUGS is positioned as optimized, zero-configuration inference microservices that expose an OpenAI-compatible API surface for serving open models inside customer infrastructure.

Date not disclosed · Official · Hugging FaceIntroducing the agentic robotics appstore for 10,000 Reachy Minis

Hugging Face publishes an appstore-style distribution concept for Reachy Mini apps/agents, tied to its open robotics ecosystem and documentation.

Date not disclosed · Official · Hugging FaceBuilding for an Open Future: partnership with Google Cloud

Hugging Face describes a partnership with Google Cloud, highlighting model deployment paths such as Vertex AI and preconfigured Hugging Face environments.

Source map · 4 recurring channels · 17 references

Funding

Latest disclosed valuation$2B
Tracked capital$164.2M6 sourced rounds

What it builds

Hugging Face Hub

A web platform where users publish and collaborate on models, datasets, and ML applications (including Spaces) with versioning and community features.

source ↗
Inference Endpoints

A managed service for deploying models to production on dedicated, autoscaling infrastructure, with deployments packaged around popular inference engines and served via an API.

source ↗
Hugging Face Inference (serverless / provider-backed inference via Hub tooling)

A unified inference approach exposed through the Hugging Face Hub libraries, including support for “Inference Providers” for serverless access to models hosted by partner providers.

source ↗
AutoTrain

A training toolchain intended to enable no-code/low-code workflows for fine-tuning models and adapting them to custom tasks/datasets.

source ↗
HUGS (Generative AI Services)

A set of optimized, zero-configuration inference microservices designed to serve open models in customers’ own infrastructure using an OpenAI-compatible API pattern.

source ↗
Spaces

Interactive app hosting on the Hub, used for running demos and community-built applications.

source ↗

Milestones & partnerships