Company

FLUX API is now available on Together AI: New FLUX1.1 [pro] and free access to FLUX.1 [schnell]

October 3, 2024

・

Together AI

We’re excited to announce support for the FLUX model family, including our fast Together Turbo endpoint for the SOTA open-source FLUX.1 [schnell], and access to the latest high-performance premium model FLUX1.1[pro]. To give developers a chance to experience the power of FLUX, we're offering 3 months of free access to FLUX.1 [schnell] as part of our partnership with Black Forest Labs (BFL), the original team behind Stable Diffusion. Through this collaboration we’re bringing SOTA image generation models to the Together Platform with the best performance, price and developer experience.

Here’s what’s now available via Together APIs and in our playground:

FLUX.1 [schnell]: A state-of-the-art open-source image generation model. We have the fastest implementation of [schnell] via our Together Turbo endpoint, with 315 ms inference time. We're also offering developers 3 months of free, unlimited API access via our FLUX.1-schnell-Free endpoint.
FLUX1.1 [pro]: BFL’s latest premium model, built for ultra-fast, high-quality image generation. Together AI is one of the exclusive launch partners for this advanced model, which delivers 3 times faster generation than FLUX.1 [pro]. Try now in our playground.‍
FLUX.1 [pro]: A premium model for high-performance use cases, accessible through our API, served in partnership with BFL. Recently improved to be 2x faster.

Free access to FLUX schnell for 3 months

Through our partnership with BFL, we're offering developers 3 months of free, unlimited API access to FLUX.1 [schnell] to experiment with image generation in their apps.

To access this offer, sign up for a free Together AI account and connect to the FLUX.1-schnell-Free endpoint. For those seeking enhanced performance and increased rate limits, you can leverage our Together Turbo FLUX.1-schnell endpoint.

Together Turbo offers the fastest performance for FLUX schnell

Our Together Turbo endpoint offers the fastest available performance for FLUX Schnell, running up to 2x faster than other implementations. Our Together Turbo endpoint can generate complex images in as little as 315 ms. Thanks to improvements from the Together Kernel Collection (TKC) and the Together Inference Engine, we enable faster image processing while maintaining quality.

For developers seeking even higher quality and faster speeds, you can access BFL’s newest FLUX1.1 [pro] model via Together AI, which generates images six times faster than FLUX.1 [pro] while also improving image quality, prompt adherence, and diversity. Try FLUX1.1 [pro] in our playground now.

Why developers ❤️ FLUX

The FLUX model has improved image generation in several areas. We've been impressed with the quality of its outputs, particularly in:

Prompt precision and output diversity

FLUX can follow complex instructions, generating detailed multi-object scenes suitable for visual storytelling and creative design projects.

*Example from BFL: two cute spiders in victorian outfits having a miniature tea party with a tiny table and teapot on a leaf, macro photo*

Text generation

FLUX handles text within images effectively, including challenging words and repeated letters.

Example from BFL: A robot holding chalk looking at a blackboard that reads the following poem:”ln pixels’ dance, AI’s craft will rise, Transforming visions through machine eyes, From dreams to screens, new worlds unfurled, AI’s brush reshapes our visual world.”

Complex compositions

FLUX excels at generating scenes with multiple, precisely positioned elements. For example, it can accurately create a complex scene with two owls having dinner, with a foreground and background, with text:

Example from BFL: underwater scene of two owls sitting at a fancy dining table with a lit candle in the center, the two owls are having a lovely dinner together. The left owl is wearing a tuxedo, the right owl is wearing a nice dress. In the background is a submarine passing by with the words “What a Hoot” painted on its side. There are little jellyfish swimming around at the bottom of the image below the table, cinematic beautiful digital artwork

Improved hand generation

While not perfect, FLUX shows significant improvement in generating realistic hands, often producing the correct number of fingers in appropriate positions.

*filmic photo of a couple on a street downtown, they are holding their hands up the camera*

Getting started

Try the FLUX models in our playground or connect via API:

Get free access to FLUX.1 [schnell] via our FLUX.1-schnell-Free endpoint (available for 3 months).

Use our Python SDK to quickly integrate FLUX models into your applications:


import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ.get("TOGETHER_API_KEY"), base_url="https://api.together.xyz/v1"
)

response = client.images.generate(
    prompt="A cat holding a sign that says hello world"",
    model="black-forest-labs/FLUX.1-schnell",
    n=1,
)

print(response.data[0].url)

‍

‍

Lower
Cost
20%
faster
training
4x
network
compression
117x

Build with FLUX.1 [schnell] for free

Experiment with the FLUX.1 [schnell] model for 3 months free via our FLUX.1-schnell-Free endpoint.

Get started

Try our super-fast FLUX.1 [schnell] Turbo

Our Together Turbo endpoint has the fastest implementation of FLUX.1 [schnell] available.

Try now

Q: Should I use the RedPajama-V2 Dataset out of the box?

RedPajama-V2 is conceptualized as a pool of data that serves as a foundation for creating high quality datasets. The dataset is thus not intended to be used out of the box and, depending on the application, data should be filtered out using the quality signals that accompany the data. With this dataset, we take the view that the optimal filtering of data is dependent on the intended use. Our goal is to provide all the signals and tooling that enables this.

Try our super-fast FLUX.1 [schnell] Turbo

Our Together Turbo endpoint has the fastest implementation of FLUX.1 [schnell] available.

Try now