⏱ 7 Reading Time
FLUX.1 Kontext is an in-context image generation and editing model from Black Forest Labs that accepts text and reference images together, applying local edits while preserving character identity and style across turns. Black Forest Labs shipped it on May 29, 2025, and it still runs in production through the BFL API, Replicate, fal.ai, Together AI, and ComfyUI.
What Is FLUX.1 Kontext?
FLUX.1 Kontext is a 12-billion-parameter rectified-flow transformer built for joint text-and-image prompting, letting a user feed a reference photo and a written instruction into one editing pass instead of a separate generation and inpainting step. Black Forest Labs, the Freiburg, Germany-based lab founded in 2023 by former Stability AI engineers Robin Rombach, Andreas Blattmann, Patrick Esser, and Dominik Lorenz, built Kontext on top of its original FLUX.1 architecture. The model distills sampling down to as few as 8 steps, which cuts inference latency instead of relying on the 20-50 step schedules older diffusion models need.
Kontext solves a specific problem text-to-image models could not: editing one region of an existing image — a shirt color, a background, a facial expression — without regenerating the entire scene from scratch. Black Forest Labs validated this with KontextBench, an internal benchmark built from crowdsourced real-world editing requests across 6 task categories, where Kontext Pro posted the top score in text editing and character preservation.
| Attribute | Value |
|---|---|
| Developer | Black Forest Labs (Freiburg, Germany) |
| Release Date | May 29, 2025 (Pro & Max); June 26, 2025 (Dev) |
| Parameters | 12 billion |
| Pricing | Pro: $0.04/image · Max: $0.08/image · Dev: free (non-commercial) / $999 per month (commercial license) |
| Platforms | BFL API & Playground, Replicate, fal.ai, Together AI, DataCrunch, Vercel AI Gateway, ComfyUI |
| Key Feature | Multi-turn, in-context editing with character and style consistency across sequential edits |
Black Forest Labs now points new projects toward FLUX.2, released in November 2025, positioning Kontext as a previous-generation model it still supports in production (source: BFL API documentation).
What Are FLUX.1 Kontext’s Key Features?
FLUX.1 Kontext ships in 3 variants — Pro, Max, and Dev — each targeting a different combination of speed, quality, and deployment control. The feature set breaks down as follows:
- Process text and reference images in a single prompt, skipping separate masking or inpainting workflows.
- Preserve character identity, facial structure, and style across multiple sequential edit turns, feeding each output back in as the next input.
- Generate at 8x lower latency than FLUX1.1 Pro, with Kontext Pro averaging roughly 7 seconds per image in independent third-party testing (source: Tigris Data benchmark).
- Apply localized edits — swapping a background, changing an object’s color, removing an item — while leaving untouched regions of the image pixel-stable.
- Deploy the open-weight Dev variant on local or self-hosted infrastructure through Hugging Face, including NVIDIA Blackwell-optimized TensorRT builds that run 2.14x faster than the base model.
- Output images at multiple aspect ratios, including 1:1, 4:3, 3:4, 16:9, 9:16, and 21:9, through the standard API request format.
Iterative editing has one documented limit: Black Forest Labs’ own demonstrations show visual artifacts and quality degradation appearing after more than 6 consecutive edit rounds on the same image (source: Black Forest Labs, FLUX.1 Kontext announcement).
How Much Does FLUX.1 Kontext Cost?
FLUX.1 Kontext runs on pay-per-image API pricing with no subscription tier: Kontext Pro costs $0.04 per image, Kontext Max costs $0.08 per image, and Kontext Dev is free for non-commercial use or $999 per month for a commercial license. Black Forest Labs’ official API documentation lists these figures directly, and Together AI confirms the same $0.04 rate for Kontext Pro on its own model card.
Third-party inference providers resell the same weights at varying markups. Replicate charges approximately $0.055 per image for Kontext Pro, while SiliconFlow prices Kontext Dev at $0.015 per image and Kontext Pro at $0.04 per image. Fal.ai matches Black Forest Labs’ official rate of $0.04 for Pro and $0.08 for Max. Some third-party gateways, including Vercel AI Gateway, issue $5 in free credits every 30 days to unpaid accounts for testing before committing to a paid workflow.
No official Black Forest Labs consumer subscription exists for Kontext specifically — separate third-party wrapper apps using the “Flux Kontext AI” name sell yearly plans starting at $6.99, but these are unaffiliated resellers layered on top of the same underlying API (source: Capterra), not Black Forest Labs products.
What Are the Pros and Cons of FLUX.1 Kontext?
FLUX.1 Kontext wins on character consistency and editing speed but loses ground to Google’s Nano Banana Pro and OpenAI’s GPT Image 1.5 on strict prompt-detail accuracy in 2026 head-to-head testing.
Pros:
- Ranks first among tested models on KontextBench for text editing and character preservation (source: Black Forest Labs technical report).
- Generates edits in roughly 7 seconds on average, the fastest of the three models tested by Tigris Data against gpt-image-1 and gemini-2.5-flash-image-preview.
- Ships an open-weight Dev variant, letting teams self-host and eliminate per-image API costs at scale.
- Holds compositional structure well during targeted local edits, according to independent testing from Artificial Analysis.
Cons:
- Loses detail-accuracy comparisons against GPT Image 1.5 and Nano Banana Pro on complex, multi-object prompts requiring exact counts or text rendering (source: felloai.com comparison, December 2025).
- Degrades visually after 6+ sequential edit turns on the same image, per Black Forest Labs’ own documentation.
- No longer Black Forest Labs’ flagship — the company now recommends FLUX.2 for new projects, meaning Kontext receives lower development priority going forward.
- Lacks real-time web search grounding that Google’s Nano Banana Pro uses to generate accurate depictions of current events or real places.
How Does FLUX.1 Kontext Compare to Google Nano Banana Pro?
Google’s Nano Banana Pro beats FLUX.1 Kontext Pro on instruction-following accuracy and real-world grounding, while Kontext Pro beats it on price and raw generation speed.
| Attribute | FLUX.1 Kontext Pro | Google Nano Banana Pro |
|---|---|---|
| Developer | Black Forest Labs | Google DeepMind |
| Price per image | $0.04 | $0.067 |
| Average generation time | ~7 seconds | 3-5 seconds |
| Underlying model | 12B-parameter flow transformer | Built on Gemini 3 Pro |
| Real-world search grounding | No | Yes, connects to Google Search |
| Best documented strength | Character preservation, local edits | Complex instruction following, typography |
Readers comparing these two models for a specific production workflow can review the full breakdown in Nano Banana Pro vs FLUX.1 Kontext: Which AI Image Editor Wins in 2026? on Knowara.
Who Should Use FLUX.1 Kontext?
FLUX.1 Kontext fits developers building character-consistent editing pipelines, indie creators who need fast local edits without a subscription, and teams that want the option to self-host through the open-weight Dev variant.
- API developers building product-photography or e-commerce tools that place the same item or model across multiple scenes benefit from Kontext’s character-preservation scores.
- Solo creators and small agencies on tight per-image budgets get a sub-$0.05 editing model with no monthly commitment.
- Enterprise teams with data-residency requirements can deploy Kontext Dev on their own infrastructure through the Non-Commercial License or the $999/month commercial license, avoiding third-party API calls entirely.
- Teams needing strict prompt-detail accuracy — exact object counts, dense typography, multilingual text rendering — get better results from GPT Image 1.5 or Nano Banana Pro instead.
What Are the Best Alternatives to FLUX.1 Kontext?
FLUX.2, Google Nano Banana Pro, and OpenAI GPT Image 1.5 are the three strongest alternatives to FLUX.1 Kontext in 2026.
- FLUX.2 is Black Forest Labs’ own successor model, released November 2025, offering multi-reference support for up to 10 images, 4-megapixel output, and improved text editing — read the full FLUX.2 Review on Knowara.
- Google Nano Banana Pro connects to Google Search for real-world accuracy and wins on complex instruction-following, at a higher $0.067-per-image rate.
- OpenAI GPT Image 1.5 leads on text rendering inside images and ecosystem integration with ChatGPT, priced at $0.034-$0.05 per image depending on quality tier.
Frequently Asked Questions
Is FLUX.1 Kontext free to use?
Kontext Dev is free for non-commercial use through Hugging Face. Kontext Pro and Max run on paid, pay-per-image API pricing starting at $0.04 per image, with no permanently free tier for commercial use.
What’s the difference between Kontext Pro, Max, and Dev?
Pro balances speed and quality at $0.04 per image, Max prioritizes prompt adherence and typography at $0.08 per image, and Dev is the open-weight version developers can self-host or fine-tune.
Can I run FLUX.1 Kontext locally?
Kontext Dev’s weights are downloadable from Hugging Face and run locally through ComfyUI or custom PyTorch pipelines, including NVIDIA Blackwell-optimized TensorRT builds for faster local inference.
Has FLUX.2 replaced FLUX.1 Kontext?
Black Forest Labs now recommends FLUX.2 for new projects and lists Kontext as a previous-generation model in its own API documentation, though Kontext Pro, Max, and Dev remain fully available and supported.