LLM tab

What the LLM tab is

The LLM tab of the Settings screen (/settings?tab=llm) is where you connect the AI that powers Testver’s assistant, test generation, recording, and failure analysis. Without a provider connected, the rest of Testver still works for manual test management — only the AI-powered features stay unavailable.

Each provider appears as an expandable card showing its name, a short description, a status indicator, and a star marking the default provider. Click a card header to expand its configuration form.

Provider	Kind	Models / role	Needs a key?
Testver	Cloud	Managed AI — a frontier model selected and optimized by Testver	Connect your Testver account on Testver → Connections; the card shows the status
Ollama	Local	Local models — Llama, Mistral, Phi, etc.	No
LM Studio	Local	Local model server	No

Testver is the only cloud provider. You do not bring your own OpenAI, Anthropic, Google, or xAI key — Testver supplies the cloud model itself, already selected and tuned for test automation work. That means no third-party model account to create, no per-vendor keys to rotate, and no decision about which model to pick.

The two local providers are unchanged: they run entirely on your own machine, cost nothing per token, and send no data anywhere.

Who sees which providers

Role	Providers shown
Any user	Testver
Administrator	Testver, plus the local providers (Ollama, LM Studio)

So if you’re not an admin and the LLM tab shows only the Testver card, that’s expected — local providers are an administrator-level configuration.

The Testver card

Expand the Testver card to see the connection status, your current plan and remaining credits — the same figures as your testver.com dashboard — plus the model settings. Connecting and disconnecting happen in one place, Testver → Connections (the pinned entry at the bottom of the sidebar); the card links there. One connection there switches on every Testver service your plan includes: AI credits, Testver Grid and Testver Mail.

Tip

For almost everyone, connecting Testver is the whole job: one step, and every AI feature in the product works. Choose a local provider instead when cost or keeping prompts on your own hardware matters more than model capability.

Provider status — the colored dot and label on each collapsed card tell you how the provider is configured:

Status	Meaning
Not Configured (gray)	No key or settings saved for this provider yet.
Configured (green)	An API key has been saved in Testver’s config.
Env Var (green)	The API key is supplied via an environment variable rather than typed into Testver.
Local (amber)	A local provider (Ollama / LM Studio) that runs models on your own machine.

Default provider — the star icon on each card header sets that provider as the workspace default. The current default shows a filled amber star and a DEFAULT badge. Clicking the star on another provider switches the default immediately.

Configuring a provider

Expanding a provider card reveals its full form. The exact fields depend on whether the provider is cloud (needs a key) or local.

Field	What it does
API Key	The secret key authorizing calls to the provider (cloud providers only). Hidden by default; use the eye toggle to reveal it. A saved key shows as dots with a (saved — click to change) hint. If the key comes from an environment variable, a note names the variable instead.
Models	The provider’s model library, shown as chips. Click the star on a chip to make it the active model; click the X to remove a model (when more than one exists).
Add model input	Type a model name and press Enter or click + to add it to the library. The placeholder suggests the provider’s default model.
Base URL	The API endpoint. Optional for cloud providers (defaults to the cloud endpoint) and required for local providers. The active and default URLs are shown beneath when relevant.
Timeout (ms)	How long to wait for a response, in milliseconds. Leave blank for the default — 60s for cloud providers, 5 min for local providers.
Test Connection	Pings the provider with the current key/URL/model and reports success or failure along with the response latency in milliseconds.
Save <provider>	Persists this provider’s API key (only if newly entered), model library, active model, base URL, and timeout — plus any pending default-provider change.

Caution

Keep your API keys secret. Treat them like passwords: never paste them into shared screenshots, chat messages, commits, or public links. Anyone with your key can spend on your account. Prefer supplying keys via environment variables on shared servers, rotate a key immediately if it is exposed, and use the eye toggle only when you need to verify what you typed.

Related

• Settings overview — the other tabs on this screen.
• First-time setup — connecting a provider during your first run.
• AI Assistant — the main consumer of whichever provider you connect.
• Skills & MCP — extend what the AI knows and can reach.