FLUX.2 Release Guide: Dates, Origins, and the FLUX.1 Lineage

FLUX.2 was announced by Black Forest Labs on November 25, 2025, as the next generation of its FLUX family. Access includes hosted services and downloadable FLUX.2 [dev] weights. Announcement, Model card

What are the key facts about FLUX.2?

When did FLUX.2 launch, and what came before it?

Black Forest Labs' announcements establish these milestones. The dates identify releases; they do not establish the current availability of every endpoint mentioned in a launch article.

Keep family names and variant names together when documenting an old workflow. A saved image labeled only "FLUX" does not tell another reader whether it came from FLUX.1 [schnell], FLUX1.1 [pro], or FLUX.2 [dev].

For an existing project, look for the exact endpoint or checkpoint name in its saved configuration. Record that identifier beside the image, prompt, and generation date before deciding what to compare with a newer release.

Use PromptZone's AI model release timeline for broader release context. Keep the vendor announcement attached to any date you put in your own project notes.

Who develops FLUX.2, and how is it related to Stable Diffusion?

Black Forest Labs develops FLUX.2. Its founding announcement describes team experience building Stable Diffusion models and introduces FLUX.1 as the company's first model suite. FLUX.1 launch

The FLUX.2 announcement explicitly places the model after FLUX.1. Use FLUX.2 by Black Forest Labs when identifying the product in a workflow, comparison, or asset record. FLUX.2 launch

Developer history and model identity answer different questions. When choosing an installation guide, match its checkpoint and component names to the model you intend to run. Do the same with service documentation and API identifiers.

For the available variants and architecture, continue to the FLUX.2 model guide. This release history provides the chronological context for that selection.

What changed from the original FLUX.1 release?

FLUX.1 launched as a text-to-image family. FLUX.2's launch describes generation and multi-reference editing, emphasizing typography and structured instructions. FLUX.1 launch, FLUX.2 launch

Compare those claims with a brief that resembles your work. For a catalog image, use a named object, viewpoint, and background. For an edit, write the requested transformation separately from the properties that must remain stable.

These are different evaluations. A successful new scene does not answer whether a product photograph survives an edit with its silhouette and label intact. Keep separate acceptance notes for each task.

What limits matter when evaluating a FLUX.2 release?

A family announcement is insufficient for checking a specific deployment. BFL's current overview distinguishes reference limits and controls by model and access route. Use the entry for your chosen variant. Model overview

The [dev] model card distinguishes the model's non-commercial license from permitted uses of generated outputs. Read the actual conditions when evaluating a deployment. Model card

Downloadable weights also require a chosen inference configuration. The model card includes a quantized example with a remote text encoder; inspect each component before describing that setup as fully local. Model card

For a comparison, save the model identifier and configuration alongside every output. Keep the creative brief stable, and record changed settings explicitly so another reviewer can understand the experiment.

How can you try FLUX.2 after checking its release history?

BFL provides a Playground and API access. The FLUX.2 online access guide covers the hosted workflow. Model overview

For a small API trial, configure a BFL key in BFL_API_KEY. This adapted request uses the documented fixed flux-2-pro endpoint. Model overview, API tutorial

curl --fail-with-body https://api.bfl.ai/v1/flux-2-pro \
  -H "x-key: $BFL_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "prompt": "Catalog photograph of a yellow desk lamp on a gray table, side view, soft daylight, plain background.",
    "width": 1024,
    "height": 1024
  }'

Poll the returned polling_url and retrieve result.sample when status is Ready. Treat Error and Failed as failures; download the image within the signed URL's ten-minute lifetime. API tutorial

For self-managed inference, begin with the [dev] model card and its implementation links. It lists ComfyUI and Diffusers support. Model card

Save the generated file with the exact request. Inspect the lamp's outline, position, and lighting against the brief, and record any corrections you would need before using the result.

What else should you know about FLUX.2's release?

What is the FLUX.2 release date?

Black Forest Labs announced FLUX.2 on November 25, 2025. That is the initial family announcement date; use a variant's own release documentation for its individual history. Announcement

Did FLUX.2 follow FLUX.1?

Yes, Black Forest Labs presents FLUX.2 as the next generation after FLUX.1. The company originally announced the FLUX.1 family on August 1, 2024. FLUX.2 launch, FLUX.1 launch

Are all FLUX.2 variants downloadable?

No, BFL distinguishes hosted models from downloadable variants. FLUX.2 [dev] is available as weights under its model-card license. Model overview, Model card

Does a fixed FLUX.2 endpoint receive preview updates?

BFL describes flux-2-pro as a fixed snapshot and flux-2-pro-preview as receiving newer improvements. Both share the same request and response contract. Model overview

Sources

• Black Forest Labs FLUX.1 announcement
• Black Forest Labs FLUX1.1 and API announcement
• Black Forest Labs FLUX.2 announcement
• Black Forest Labs FLUX.2 model overview
• Black Forest Labs FLUX.2 dev model card
• Black Forest Labs text-to-image API tutorial
• ByteDance Seedream 4.0 product page
• BytePlus image generation tutorial

Related guides on PromptZone

• Realistic Photos with FLUX
• Best SDXL Models in 2026