Overview
Publish Owl supports multiple AI providers for both content generation and image creation. Each provider has different strengths, pricing, and capabilities. This guide helps you choose the right provider for your workflow.
Add your API keys in Settings. You can use different providers for different workflow steps within the same workflow.
Content Generation Providers
| Provider | Default Model | Web Search | Best For |
|---|---|---|---|
| OpenAI | gpt-5-mini | Yes (+$0.01) | General purpose, strong instruction following |
| Anthropic | claude-sonnet-4-5 | Yes (+$0.01) | Nuanced writing, long-form content |
| Grok (xAI) | grok-4-1-fast-reasoning | Yes (+$0.005) | Cheapest web search, X/Twitter integration |
| Gemini | gemini-2.5-flash | Yes (+$0.035) | Fast, cost-effective, Google search integration |
| OpenRouter | (varies) | No | Access to 100+ models through one API key |
| Perplexity | sonar | Always on | Research-first, every response includes web data |
| Replicate | meta/meta-llama-3.1-405b-instruct | No | Open-source models, custom model support |
You can enter any model ID in the model field — these defaults are just starting points. Check each provider's documentation for available models.
Image Generation Providers
| Provider | Default Model | Cost/Image | Key Features |
|---|---|---|---|
| OpenAI | gpt-image-1.5 | $0.01-0.17 | Reference images for style transfer, transparent backgrounds, multiple output formats |
| Gemini | gemini-2.5-flash-image | $0.02-0.24 | Google's image generation |
| Flux | flux-2-pro | $0.03-0.08 | Multiple tiers (Pro, Max, Ultra), "Raw" mode for natural aesthetics |
| Stable Diffusion | sd3.5-large | $0.01-0.08 | 40+ style presets (photographic, cinematic, anime, etc.) |
| Ideogram | V_3 | $0.02-0.09 | Style types (Realistic, Design, Fiction), color palettes, Magic Prompt |
| xAI Grok | grok-imagine-image-pro | $0.07 (pro) / $0.02 (standard) | xAI's image generation models |
| Replicate | flux-schnell | $0.003-0.07 | Cheapest option, open-source models, custom model IDs |
Stock Photo Providers
All stock photo providers are free to use:
- Pexels: Unlimited requests. Large library of professional photos.
- Pixabay: 100 requests per minute limit. Photos, illustrations, and vectors.
- Unsplash: High-quality artistic photography. Use the Access Key (not Secret Key) from your developer account.
Replicate Model Library
Replicate provides access to popular open-source models. Pre-configured options include:
| Model | Cost/Image |
|---|---|
flux-schnell (default) | ~$0.003 |
flux-dev | ~$0.025 |
flux-pro | ~$0.055 |
flux-2-max | ~$0.07 |
flux-kontext-pro | ~$0.04 |
sdxl | ~$0.018 |
stable-diffusion-3.5-large | ~$0.065 |
ideogram-v3-turbo | ~$0.02 |
ideogram-v3 | ~$0.04 |
You can also enter any Replicate model ID for both text and image generation. Browse available models at replicate.com/explore.
Choosing the Right Provider
For research-heavy content
Use Perplexity or enable web search on OpenAI/Grok for the research step. Perplexity is inherently search-based; Grok offers the cheapest per-search pricing.
For budget-conscious image generation
Replicate's flux-schnell at ~$0.003/image is the cheapest AI option. For free images, use Pexels, Pixabay, or Unsplash.
For highest quality images
GPT Image 1.5 (OpenAI) is recommended for its reference image support and consistent quality. Flux Pro/Max and Ideogram V3 also produce excellent results.
For multi-step workflows
Mix providers across steps. For example: Perplexity for research, Claude for writing, GPT for editing. Each step can use a different provider and model.