AI Providers & Models

B-Roll Me supports four AI providers. You can use any combination — the app fetches available models from all providers that have a valid API key configured.

Using Credits mode?

You don't need to configure any of the providers below — the app handles AI calls automatically using your credit balance. Model selection is also handled for you. See Credits & Pricing for details.

The providers below are only needed if you're using BYOK (Bring Your Own Key) mode. In BYOK mode, you supply your own API keys and choose which models to use for each task. See Authentication for how to switch between modes.

Supported Providers

Anthropic (Claude)

Models like Claude Sonnet, Claude Haiku, and Claude Opus. The app filters to models with IDs starting with claude-.

Sign up: console.anthropic.com · Strengths: Excellent analysis quality, strong instruction following

OpenAI (GPT)

Models like GPT, GPT Mini, and Codex. The app excludes instruct, realtime, audio, and transcription variants.

Sign up: platform.openai.com · Strengths: Fast, widely used, good cost/performance ratio with GPT mini

Google Gemini

Models like Gemini Flash and Gemini Pro. Filtered to models that support content generation.

Sign up: aistudio.google.com · Strengths: Very cost-effective, generous free tier for experimentation

OpenRouter

Access to 100+ models from multiple providers through a single API. Model listing is always available (no key required for browsing).

Sign up: openrouter.ai · Strengths: Maximum model variety, unified billing, useful for trying different models

Model Selection

In Credits mode, model selection is handled automatically — you don't need to configure anything. In BYOK mode, Settings > Analysis Preferences shows a model selector with all models available from your configured providers. Models are displayed as provider:model_id strings, for example:

  • anthropic:claude-sonnet-4-6
  • openai:gpt-5
  • gemini:gemini-3-flash-preview
  • openrouter:deepseek/deepseek-r1

Per-Feature Model Overrides

B-Roll Me lets you assign different models to different tasks:

Script Analysis

Identifies B-Roll moments from your script. Use a fast, affordable model here — GPT Mini or Gemini Flash work well.

Clip Evaluation

Scores clips for relevance. Use a more capable model here if you want higher-quality scoring — Claude Sonnet or GPT 5 are good choices.

To set an override, enable the toggle for the feature in Settings > Analysis Preferences and select a model from the dropdown. When disabled, the feature uses the default model.

Cost Optimization Tips

  • Use GPT Mini or Gemini Flash for script analysis (fast, cheap, good enough for identifying moments).
  • Use Claude Sonnet or GPT 5 for clip evaluation (quality matters more for accurate scoring).
  • Reduce the max moments setting if your scripts are long — fewer moments means fewer API calls during search and evaluation.
  • OpenRouter is useful for comparing models without switching providers.