flux.1-kontext-max

by bfl.ai

Pricing

Input $0.00 / 1M tokens
Output $0.03 / 1M tokens

FLUX.1 Kontext [Max] by Black Forest Labs


Overview

FLUX.1 Kontext [Max] is a 30-billion parameter advanced multimodal image editing model from Black Forest Labs, extending the FLUX.1 Kontext series with enhanced capacity and performance. This proprietary model supports complex in-context editing tasks with text and image inputs, delivering the highest fidelity and flexibility for professional and enterprise-grade workflows. It is optimized for large-scale deployment with NVIDIA RTX GPUs and TensorRT acceleration.


Features

  • Enhanced multimodal in-context editing combining text and image inputs
  • Superior text editing, character consistency, and style transfer
  • Optimized for fast inference on NVIDIA RTX GPUs via TensorRT
  • Enterprise-grade API access and licensing for scalable integration
  • Supports larger context windows and more complex workflows

Use Cases

  • Complex professional image editing with detailed context and style control
  • Maintaining consistent character and style across extended projects
  • Enterprise applications requiring high-fidelity, large-scale image generation
  • Advanced context-aware style transfer and iterative image modification

Limitations

  • Proprietary model with commercial licensing requirements
  • Requires NVIDIA RTX GPUs and TensorRT for optimal inference performance
  • No public weights available; access only via API and licensed deployments
  • High resource requirements due to large model size and complexity

Access & Integration

  • Available through Black Forest Labs API for commercial and enterprise use
  • Integrated on platforms such as RunComfy and Replicate
  • Optimized for NVIDIA RTX GPUs with TensorRT backend for scalable performance

Quick Comparison

Feature FLUX.1 Kontext [Pro] FLUX.1 Kontext [Max]
Parameters 12 Billion 30 Billion
Weight Access Proprietary, commercial license Proprietary, commercial license
Input Modalities Text + Image Text + Image
Output Quality Professional-grade, high fidelity Highest fidelity, most complex edits
Hardware Optimization RTX + TensorRT optimized RTX + TensorRT optimized
Commercial Use Yes Yes