Model Selection

Key features

• System Default — Use the workspace’s configured model
• Provider list — OpenAI, Anthropic, Gemini, DeepSeek, xAI, Perplexity, Mistral, Kimi, and others
• Model list — All models for the selected provider
• Capability icons — Reasoning, Speed, Code, Search, Vision, Chat
• Pricing display — Credits per million tokens when available
• Plan enforcement — Some models may require Premium; those show an upgrade prompt
• Mobile layout — Providers and models shown as a two-step flow

Opening the modal

1. In the chat input toolbar, find the Model: label.
2. Click the label or the dropdown arrow.
3. The modal opens above the input.

Close by clicking outside, pressing Escape, or choosing a model.

Left column: providers

System Default

1. First option at the top
2. Uses the workspace’s system LLM preference
3. Shows the actual default model name when known (e.g. "System Default (gpt-4)")
4. Selecting it clears any per-thread override

Provider list

Providers are shown with:

1. Logo — Provider icon
2. Name — Provider label (e.g. OpenAI, Anthropic)
3. Model count — Number of models for that provider

Providers may be reordered or hidden based on instance configuration. Disabled providers are not shown.

Right column: models

After selecting a provider, the right side shows its models.

Model row contents

| Element | Description | |--------|-------------| | Model name | Display name (e.g. "GPT-4", "Claude 3 Opus") | | Capability icons | Tags for reasoning, speed, code, search, vision, chat | | Pricing | Credits per million tokens (input+output average) when available | | Premium badge | Shown when your plan does not allow that model |

Capability icons

| Icon | Capability | Meaning | |------|------------|---------| | Brain | Advanced Reasoning | Better for complex analysis | | Lightning | Fast Response | Optimized for speed | | Code | Code Specialist | Strong for coding tasks | | Magnifying glass | Web Search | Supports web search | | Eye | Vision & Images | Can process images | | Chat | Chat | General conversation |

Provider “Best for” tags

Each provider may show suggested use cases (e.g. General, Analysis, Creative, Code, Search, Long Context). These help you pick a provider for your task.

Selecting a model

1. In the left column, click a provider.
2. Wait for models to load.
3. In the right column, click a model.
4. The modal closes and the selection is applied to the current thread.

The Model: button updates to show the selected model name. If the provider has connection issues, an error message appears instead of the list.

Plan limits

1. Allowed models — Models available to your plan can be selected.
2. Restricted models — Show a “Premium plan” badge and cannot be selected; an upgrade prompt may appear if you try.
3. Capabilities — If plan data is unavailable, selection may be limited.

Persistence and scope

1. Thread-scoped — The chosen model is stored per thread in localStorage.
2. No thread — If you’re not in a thread (e.g. home), the selection applies to the next thread you open.
3. System Default — Removes the per-thread override; the workspace default is used again.

Mobile behavior

On mobile:

1. First view — Providers only.
2. Select provider — Switches to the models list.
3. Back — Returns to the providers list.
4. Close — X button or tap outside.

Troubleshooting

1. Models not loading — Check provider configuration and API keys in Instance Settings.
2. "Connection Error" — API unreachable or misconfigured; verify provider settings.
3. Model greyed out — Likely restricted by plan; upgrade to access it.
4. Selection not sticking — Ensure you’re in a thread; try selecting again.
5. System Default empty — Workspace LLM may be unset; configure it in Workspace Settings.

Related topics

For more information, see:

1. Chat Input Field Guide — Where the model selector lives
2. Chat Agent Selection Guide — Choosing agents
3. Workspace Settings: Chat Settings — Workspace default model