recursal.ai Open in urlscan Pro
2606:4700:20::681a:450  Public Scan

Form analysis
0 forms found in the DOM

Text Content

Why Recursal

How it works

Pricing

Blog

Login

Try Recursal


RECURSIVELY BUILDING THE NEXT AI MODEL, FOR EVERYONE, IN EVERY LANGUAGE

Try Recursal




(WHY RECURSAL)


AFFORDABLE

Our linear transformer architecture is orders of magnitude more scalable then
existing transformer based architecture. Allowing us to run the same workload of
the biggest AI provider, at a 1/100th of the infrastructure footprint. (It also
makes us the world’s greenest AI model)

Find out more


MULTILINGUAL

Trained on over a 100 languages, with a custom multilingual tokenizer, we build
our AI model for global usage - not just the 17%, consisting of the English
speaking world.

Find out more


OPEN SOURCE

Our team came together from the Eagle / RWKV open source community project. We
built this platform, to scale both the usage of our open source model, and
finance the recursive development of all future models, within the open source
space.

Find out more


(HOW IT WORKS)


UNLOCKING THE

MECHANICS OF PROCESS:

Step 1


REGISTER, AND SETUP RECURSAL AI BETWEEN YOU AND YOUR EXISTING AI MODELS

 If our AI isn't equipped to handle a query, it seamlessly delegates it to your
existing model, while Recursal continually learns and adapts, providing some
answers independent of GPT over time.

Step 2


AUTOMATE DATA COLLECTION, AND TRAINING OF YOUR OWN MODEL. REMOVE YOUR DEPENDENCY
ON EXISTING PROPRIETARY MODELS OVER TIME.

Enjoy the benefits of both Recursal AI’s affordability for common tasks, while
still routing complex use cases to the larger main model. Ensuring the best
solution for every scenario

Try Recursal



(Pricing)


START SAVING WITH RECURSAL

Flexi


FREE

Per month

Inference tokens*

0.5M

Storage limits

/ limit

100GB

Start with Flexi

* only on Eagle 7B models

Pro


$20

/month

Per month

Inference tokens*

200M

Finetune tokens

20M

Router requests

10M

Storage limits

/ limit

500GB

Start with Pro

* 200M is only for Eagle 7B models, 50M tokens when using the Mixtral model

On demand

Inference Cost (Eagle 7B)

/ 1M tokens

$0.20

Inference Cost (Mixtral)

/ 1M tokens

$0.80

$0.10

$0.40

Additional

Additional Finetune ( Eagle 7B )

/ 1M tokens



Additional Router Request

/ 1M requests



$1

$2


(JOIN US)


START SAVING WITH RECURSAL

Over time you can completely exclude GPT from the equation and save up 10 times
the price of GPT

Get started

Blog

Terms of service

Privacy

Subscribe not to miss Recursal AI updates




© 2024 Recursal. All rights reserved.

Designed by 0o