# FLUX 1 Kontext Dev

> FLUX 1 Kontext Dev is available on Artifio.ai — a unified AI workspace with a single pay-as-you-go wallet.

- Canonical page: https://artifio.ai/models/bfl-flux-1-kontext-dev
- Run it: https://artifio.ai/create/bfl-flux-1-kontext-dev
- Modality: image editing
- Model brand: Black Forest Labs
- Pricing: from about $0.015 per generation (exact cost shown in the workspace before you confirm; varies with selected options)
- Typical generation time: ~30s (catalogue estimate)
- Full pricing table: https://artifio.ai/pricing/models

## About FLUX 1 Kontext Dev

FLUX 1 Kontext Dev is the open-weight member of Black Forest Labs' FLUX.1 Kontext suite, a family of flow-matching models announced on 29 May 2025 that edit images from a text instruction plus a reference image. The dev variant, a 12-billion-parameter diffusion transformer, spent a month in private beta for research and safety testing before its public weight release on 26 June 2025 under the FLUX.1 non-commercial license.

Unlike its pro and max siblings, Kontext Dev concentrates entirely on editing rather than generation from scratch. On Artifio it runs alongside 100+ other models in one workspace, paid per generation from a single wallet.

## Features

- 12-billion-parameter open-weight diffusion transformer specialized in instruction-based image editing
- Performs both precise local edits and global changes from a text instruction and an input image
- Preserves characters across a diverse set of scenes and environments through iterative, multi-turn editing
- Ships in BF16, FP8 and FP4 TensorRT variants developed with NVIDIA for faster inference and lower memory use on Blackwell GPUs
- Weights are free for research and non-commercial use, with commercial licenses available through a self-serve portal

## What people use it for

- Swapping or restyling a single object in a photo without regenerating the rest of the frame
- Keeping a character's appearance consistent across a series of edited scenes
- Editing or replacing text that appears inside an image
- Research on instruction-based editing in self-hosted pipelines where open weights are required

## Reported strengths

- In human-preference evaluations on KontextBench it beat open editing models such as ByteDance Bagel and HiDream-E1-Full, and Google's Gemini Flash Image in many categories, results Black Forest Labs says were confirmed independently by Artificial Analysis
- The technical report highlights strong preservation of objects and characters, which makes iterative editing workflows unusually robust
- TensorRT builds give it practical speed and memory options on recent NVIDIA hardware

## Reported limitations

- It is an editing model only; it does not replace a general text-to-image generator
- Black Forest Labs documents visual artifacts that accumulate after excessive multi-turn editing sessions
- The model can fail to follow instructions and has limited world knowledge, and distillation can introduce visual artifacts
- The weights are non-commercial; commercial self-hosting requires a paid license

## Sources

1. FLUX.1 Kontext [dev] open weights release (Black Forest Labs): https://bfl.ai/announcements/flux-1-kontext-dev
2. Introducing FLUX.1 Kontext and the BFL Playground (Black Forest Labs): https://bfl.ai/announcements/flux-1-kontext
3. FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space (arXiv): https://arxiv.org/abs/2506.15742

---

About Artifio.ai: 100+ AI image, video, and audio models in one workspace with a single shared wallet — no per-model subscriptions. This is the markdown rendition of https://artifio.ai/models/bfl-flux-1-kontext-dev for AI assistants and crawlers; the HTML page is canonical.