Harmony
AI Models

Access 1,800+ AI models
from 70+ providers

Choose from leading AI providers and models, all available inside Harmony. No need to switch tools; pick the right model for Companion, insights, and workflows.

Trusted by teams at
DocplannerRTB HouseTotvsYOU BPOTECHFluencyNetguru
How it works

Pick a model. Swap it any time.

Step 01

Choose the right model

Filter by capability. Vision, tool use, reasoning, context window. Pick per workspace, per Skill, or per Agent.

Step 02

Use it everywhere in Harmony

The same model powers Companion, Skills, Agents, and Studio. One choice, consistent behavior across features.

Step 03

Swap without rewriting

Change providers when pricing, latency, or quality shifts. Prompts, tools, and guardrails come along untouched.

Model picker

The best model for the moment. Not the loudest one.

For a long-context recap you might reach for Claude Sonnet. For a fast agent hop, Groq. For a private deployment, Azure OpenAI. Harmony lets you pick per surface without rewiring anything.

Provider-agnostic. No lock-in, no rewrites. Capability filters surface the right model for the job. Enterprise workspaces bring their own keys and endpoints.
Model picker
For · Companion · Sales workspace
Filters · vision · tools
Claude Sonnet 4
Anthropic · 200K context
toolsreasoningvision
GPT-5.1
OpenAI · 400K context
toolsreasoningvision
Gemini 2.5 Pro
Google · 1M context
toolsreasoningvision
Llama 3.3 70B
Groq · 128K context
tools
Applies to Companion · Skills · Studio
Swap any time. No prompts to rewrite.
Supported Providers

Leading AI providers, ready to use

Use the same models you already trust, grouped by provider, available in your workspace.

Vision Tools Reasoning
Capabilities

One toolkit. Every provider.

Capability filters

Sort by vision, tool use, reasoning, and context window. Pick the right model without a spreadsheet.

Per-surface model routing

Companion, Skills, Agents, and Studio can each use a different model. Match the model to the task.

Bring your own keys

Point workspaces at your own OpenAI, Anthropic, Azure, or Bedrock accounts. Billing and data stay with you.

Private endpoints

Self-hosted models and VPC-scoped endpoints for regulated deployments. No traffic leaves your boundary.

Streaming by default

First-token latency stays under a second for supported providers, so Companion feels live end to end.

Usage and cost observability

Track spend per model, per team, per feature. Set caps, get alerts, catch runaway usage early.

Enterprise

Bring your own models

Connect your own LLM infrastructure. OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, or self-hosted open-source models. Your API keys, your endpoints, your data boundaries.

Learn more
The model landscape shifts weekly

Your stack should not have to. Pick, ship, and swap when the market moves.

Ready to use your
favorite models?

Connect the AI providers your team already trusts
and switch between them without ever leaving Harmony.