> ## Documentation Index
> Fetch the complete documentation index at: https://www.recraft.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# External models

External models let you utilize top AI models directly in Recraft Studio, so you can generate and edit images without switching tools.

| Model                  | Creator           | Description                                                                                               | Time | Credits |
| ---------------------- | ----------------- | --------------------------------------------------------------------------------------------------------- | ---- | ------- |
| Flux 2 Max             | Black Forest Labs | Maximum consistency for complex, multi-image workflows                                                    | 20s  | 18      |
| Flux 2 Dev             | Black Forest Labs | Open-weight model for fast experimentation and pipelines                                                  | 6s   | 3       |
| Flux 2 Pro             | Black Forest Labs | Production-grade realism with strong prompt alignment                                                     | 10s  | 10      |
| Flux 2 Flex            | Black Forest Labs | Configurable quality, speed, and typography control                                                       | 20s  | 13      |
| Flux 1 Dev             | Black Forest Labs | Developer-focused version for experimentation                                                             | 9s   | 7       |
| Flux 1.1 Pro           | Black Forest Labs | Refined Flux model with high-fidelity rendering                                                           | 6s   | 8       |
| Flux 1 Kontext Pro\*   | Black Forest Labs | Context-aware Flux variant preserving semantic alignment in scenes                                        | 7s   | 8       |
| Flux 1 Kontext Max\*   | Black Forest Labs | Advanced Kontext version with enhanced resolution and cinematic detail                                    | 12s  | 15      |
| GTP Image 1.5 Low\*    | OpenAI            | Fast, lightweight layout-aware image generation                                                           | 15s  | 2       |
| GTP Image 1.5 Medium\* | OpenAI            | Balanced quality and consistency for general use                                                          | 25s  | 5       |
| GTP Image 1.5 High\*   | OpenAI            | High-detail images with precise layout control                                                            | 50s  | 24      |
| GPT Image 1 Low\*      | OpenAI            | Lightweight GPT-4o variant for text-to-image tasks, multi image editing                                   | 15s  | 2       |
| GPT Image 1 Medium\*   | OpenAI            | Balanced GPT-4o variant with improved detail and realism, multi image editing                             | 35s  | 8       |
| GPT Image High\*       | OpenAI            | High-quality GPT-4o setting delivering photorealistic outputs, multi image editing                        | 60s  | 30      |
| HiDream I1 Dev         | HiDream AI        | Experimental diffusion model optimized for artistic compositions                                          | 12s  | 6       |
| Ideogram V3 Turbo      | Ideogram AI       | Fast inference mode optimized for speed                                                                   | 9s   | 6       |
| Ideogram V3 Default    | Ideogram AI       | Standard setting balancing speed and quality                                                              | 12s  | 11      |
| Ideogram V3 Quality    | Ideogram AI       | Highest-fidelity Ideogram variant                                                                         | 12s  | 17      |
| Imagen 4               | Google DeepMind   | Advanced Imagen with higher realism                                                                       | 8s   | 8       |
| Imagen 4 Ultra         | Google DeepMind   | Top-tier Imagen for ultra-photorealistic rendering                                                        | 10s  | 13      |
| Nano Banana\*          | Google Gemini     | Standard-resolution outputs, reliable generation, and foundational editing.                               | 12s  | 7       |
| Nano Banana Pro\*      | Google Gemini     | Faster generation, higher fidelity, accurate multilingual text, and advanced creative controls            | 15s  | 32      |
| Nano Banana Pro 4K\*   | Google Gemini     | High-resolution variant delivering sharper detail and 4K-grade image outputs.                             | 30s  | 63      |
| Seedream V4.5\*        | ByteDance         | Structured, reference-driven high-fidelity image generation                                               | 30s  | 8       |
| Seedream V4\*          | ByteDance         | Next-gen diffusion model delivering vivid colors, fine detail, and artistic realism                       | 20s  | 7       |
| Qwen-Image\*           | Alibaba           | Multimodal model built for high-fidelity image generation and text–image alignment                        | 7s   | 4       |
| Z-Image Turbo          | Alibaba           | Ultra-fast photorealism with strong bilingual (English/Chinese) text rendering and solid prompt adherence | 3    | 1       |

\*Used for [editing images with natural language](https://www.recraft.ai/docs/recraft-studio/image-editing/editing-images-with-natural-language) within Recraft.

## **FAQ**

**Why do different models have different credit prices?**

Different models have different credit prices because each model has unique capabilities and API costs. More advanced or higher-quality models require more credits per use. For example, Nano Banana Pro 4K costs more credits than basic raster generation. The credit cost reflects the model’s complexity and external API pricing.

You can find more details about how credits work [here](https://www.recraft.ai/docs/plans-and-billing/credits).