Workflow Recipes
Which AI Provider Can Do What?
What it does: One page to answer "why is this feature greyed out / erroring?" — every feature × every provider, plus how to combine providers smartly.
Applies to Antradus AI Lite, Publisher & Studio v2.7.0 · Last updated 2026-08-26
The matrix
| Capability | OpenAI | Anthropic (Claude) | Google Gemini | OpenRouter | DeepSeek |
|---|---|---|---|---|---|
| Article writing | ✅ | ✅ | ✅ | ✅ 200+ models | ✅ |
| Image generation (featured + albums) | ✅ gpt-image / DALL·E | ❌ no image API | ✅ Imagen | ✅ Flux & compatible | ❌ |
| Vision (product-from-photo) | ✅ | ✅ | ✅ | ⚠️ model-dependent | ❌ |
| Live web search | ⚠️ search-capable models only (gpt-5 / gpt-4o) | ✅ | ✅ Google grounding | ✅ any model | ❌ |
| Free tier models | ❌ | ❌ | ❌ | ✅ ★ FREE labels | ❌ |
| In-editor assistant / transforms | ✅ | ✅ | ✅ | ✅ | ✅ |
| Bulk publishing | ✅ | ✅ | ✅ | ✅ | ✅ (no web search, no images) |
Practical pairings
- "I want everything from one key" → OpenAI or Gemini (writing + images + vision + search).
- "Best writing, images secondary" → Claude for text; note you’ll have no image generation — pair with a second provider when you need images.
- "Minimum cost" → OpenRouter with a ★ FREE model for writing; expect variable quality, and remember free-model providers often retain/train on prompts.
- "Maximum privacy of drafts" → any paid API tier (paid API traffic is generally excluded from training by the major providers — check your provider’s current terms).
Switching providers
Keys are stored per provider — switching the Active Provider in settings is instant and non-destructive. A common pro setup: bulk-write a batch on a cheap/free model, then regenerate the most important articles on a flagship model.
Where the model choice shows up
Every generation uses whatever Model / Image Model is selected at that moment — articles, meta, transforms, bulk jobs, images. The Reporting page records usage per model so you can compare real costs.