The Models Hub is your central place to discover, configure, and manage the AI models that power your agents.

What you can do here

Browse models. Search across 1,700+ models from OpenAI, Anthropic, Google, Mistral, DeepSeek, and more. Filter by provider, context length, capabilities (vision, code, reasoning), or cost. Compare pricing. See the cost per million tokens for each model, so you can make an informed choice between capability and cost. Set a default model. Choose which model your agents use unless you’ve configured something different at the agent level. Configure per-agent models. Different agents can run on different models — your Developer agent might run on a strong reasoning model, while your Email Agent uses something faster and cheaper for routine triage.

Connecting providers

Models are only available after you’ve added an API key for their provider. Go to Settings → Provider Keys to add your keys. Once a provider key is active, all of that provider’s models appear in the Models Hub and are available to assign to agents. Every task type has a different best model:
TaskWhat to look forExamples
Complex reasoning, codeHigh capability, large contextClaude Opus, GPT-4o, Gemini Ultra
Research, writingGood at synthesis and proseClaude Sonnet, GPT-4o-mini
Quick responses, triageFast, low costClaude Haiku, GPT-4o-mini, Gemini Flash
Local / privateRuns on your own hardwareOllama models (Llama, Qwen, Mistral)

Smart routing

You can enable smart routing to let Qorven automatically pick the right model for each task. When enabled, Qorven considers the task type, the conversation history length, and the cost profile of available models, then selects the best fit — without you having to think about it. Smart routing works best when you have keys for multiple providers active.

Local models with Ollama

Qorven works with Ollama — run models like Llama, Qwen, and Mistral on your own hardware. Add http://localhost:11434 as your Ollama endpoint in Settings → Provider Keys, and all your Ollama models appear in the Hub. Local models never send data to any external service — useful when privacy is a priority.

Where next

Add provider keys

Connect your API keys to unlock models.

Per-agent models

Set a different model for each agent.

Local models

Run Ollama or LM Studio models on your own hardware.

Provider reference

Setup guides for each supported provider.