Skip to content
TrustList
F3
AI Model

FLUX 3 Action

Black Forest Labs' 7B open-weight model that turns camera views, joint positions and an instruction into robot-arm actions

About FLUX 3 Action

Robot arms need a policy that turns what a camera sees into motor commands. FLUX 3 Action is Black Forest Labs' attempt to supply one from the same family of models that powers its FLUX 3 image and video work.

From pixels and an instruction to motion

The model has 7 billion parameters and is described as a world action model. It adapts the multimodal backbone of FLUX 3. Its inputs are camera images, the robot's joint positions and an instruction in plain language. From these it predicts a brief run of motor commands, plus frames that picture what should happen next. The weights are open and published on Hugging Face, and VentureBeat reported the release on 23 September 2026.

What is in the release

The release has a shared base and two fine-tuned policies from Black Forest Labs. One targets the Franka arm and was trained on the DROID dataset. The other targets the low-cost SO-101 arm used with the LeRobot library. Fine-tuning code is included. Running the Franka policy takes roughly 32 GB of GPU memory at BF16 precision.

Reported results

On NVIDIA's RoboLab-120 simulation benchmark the company reports 42.92 per cent task success (its page shows 42.2 per cent for a distilled variant), and a partner's real-robot test completed 28 of 30 tasks. Both sets of figures come from the developer itself, not independent testers.

Licence terms to read first

Under the FLUX Kommunity License v1.0, commercial use is free only for organisations below a stated annual revenue threshold, so a larger company needs a separate agreement. Military, surveillance and biometric uses are ruled out. At the time of reading, no hosted API or price had been announced, which leaves self-hosting as the route for anyone wanting to try it.

TrustList summary

FLUX 3 Action is a robot-control model with 7 billion parameters, released with open weights on Hugging Face by Black Forest Labs. It adapts the backbone of the company's FLUX 3 image and video model, so that it predicts robot motion from camera images, joint positions and a plain-language instruction.

Main features

  • Motor commands: outputs a short sequence of motor commands for the robot arm.
  • Expected outcomes: also returns frames showing what the robot is expected to do next.
  • Franka policy: a fine-tuned policy for the Franka arm, trained on the DROID dataset, needing about 32 GB of GPU memory in BF16.
  • SO-101 policy: a fine-tuned policy aimed at the low-cost SO-101 arm, with code for the LeRobot library.
  • Fine-tuning code: code for adapting the shared base model is included.

Who it works with

The published policies target the Franka arm and the low-cost SO-101 arm, the two setups the developer has released.

Key facts

  • Licence: FLUX Kommunity License v1.0, which limits commercial use to organisations under a revenue threshold. Larger companies need a separate agreement.
  • Restricted uses: military, surveillance and biometric uses are prohibited.
  • Developer's results: 42.92 per cent task success on the RoboLab-120 simulation benchmark, and 28 of 30 tasks finished in a partner's real-robot test, both reported by the developer.
  • Hosted access: no hosted API announced.

Summary written by TrustList from the business's own published information, Nov 2026.

Benchmarks & AI stats

BenchmarkOfficialCommunity avg
No benchmark scores yet. Be the first to add one.

“Official” values are editor-approved and feed the ranking. “Community avg” is the mean of member submissions (shown for transparency; it never affects the ranking until an editor approves a value).

Sign in to add a benchmark score for this model.

FLUX 3 Action launches

Request a demo or quote from FLUX 3 Action

Protected by reCAPTCHA - Google Privacy Policy and Terms apply.

By sending, you agree we may share your request and contact details with the provider once you confirm your email.

Reviews

Write the first review of FLUX 3 Action

Used it? Your experience helps other buyers decide.

Write a review

Questions & answers

No questions yet. Be the first to ask about FLUX 3 Action.

FLUX 3 Action alternatives

Similar listings buyers compare against this one.

Open-weight translation model from Bilibili's Index team, covering 150 languages and following glossary and style instructions

Microsoft AI's real-time speech-to-text model for voice agents, transcribing 60 languages with continuous language detection

Black Forest Labs' image generation and editing model with bounding-box layouts, multiple reference images and 4K output

Google's open multimodal embedding model that maps text, images, audio and video into one vector space, under Apache 2.0

Open-source 2B decision model from AWS's Strands Agents project that picks options and returns scores with a reliability figure

Aleph Alpha's open-weight English-German mixture-of-experts model for self-hosting, released under Apache 2.0 in October 2026

Perplexity's open-weight decision model that picks among given options for routing and checks in AI systems

Microsoft AI's text-to-speech model for keeping one voice across 23 languages, with voice cloning from short audio