Theworld’sbestopensourcetext&imagemodels,decentralised.
Subnet 19 is a pivotal node within the Bittensor network, focused on providing access to cutting-edge open-source machine learning models. It serves as a gateway for users to tap into a diverse array of text & image models. At the core of Subnet 19 is our fully decentralised philosophy, where validators and miners run all their own models, completely decentralising the core infrastructure. We are deeply integrated in the bittensor ecosystem, and also support models from other subnets.
LLaMA 3 is a trailblazing language model acclaimed for generating human-like dialogue and written content. Its advanced capabilities merge contextual understanding and creative expression, making it a sought-after solution in customer service, content creation, and education. Built on a robust transformer architecture, LLaMA 3 empowers users to craft compelling narratives with ease and accuracy
Try out on Corcel Playground
Proteus is a state-of-the-art high dynamic range (HDR) image generation model that sets the gold standard for producing spectacular, photorealistic visuals from simple text prompts. Its cutting-edge architecture and extensive training on diverse HDR imagery allow it to masterfully capture intricate details, lifelike textures, and accurate lighting in generated scenes. From breathtaking landscapes to product renderings, Proteus consistently delivers results that blur the line between digital creation and reality.
Try out on Corcel Playground
Playground is a state-of-the-art diffusion-based AI model that excels in transforming written descriptions into stunning, high-resolution images. With the capability to generate visuals in both portrait and landscape orientations, it stands out for its versatility and the high aesthetic quality of its outputs. This model harnesses the power of advanced neural networks to interpret textual input with remarkable accuracy, producing images at resolutions of up to 1024x1024 pixels
Try out on Corcel Playground
Mixtral-8x7B is a trailblazing large language model that's capturing the spotlight for its extraordinary capabilities. With a flair for mastering the subtleties of language and an unparalleled prowess in processing intricate data, Mixtral-8x7B operates with a level of finesse that sets a new benchmark for AI excellence
Try out on Corcel Playground
Dreamshaper is a versatile text-to-image model renowned for generating high-quality, dreamlike visuals from text prompts. Its fine-tuned capabilities bridge photorealism and digital art, particularly in portrait illustrations, making it a favorite in industries like gaming, cinema, and design. Built on the robust SDXL framework with adaptable checkpoints, Dreamshaper enables users to effortlessly create everything from video game characters to classical-style paintings
Finetuned Stable Diffusion Model
BringingGenerativeAItothemasses.
Proudly powered by the Bittensor networkSubnet 19 provides access to advanced language models and generative AI. By harnessing the collective power of the Bittensor network and open-source models, Subnet 19 enables a free suite of AI-powered applications. These include engaging chat, personalized avatars, and creative image generation - putting the power of cutting-edge AI into the hands of everyone. Subnet 19's mission is to make generative AI accessible, open, and beneficial to all. At the core of Subnet 19's mission is Corcel Research Labs, a hub of innovation where development of state-of-the-art models takes place. This continuous pursuit of excellence in AI research ensures that Subnet 19 remains at the forefront of decentralised machine learning services.
Proudly powered by the Bittensor network01
Unmatched Reliability
Subnet 19, with its robust network of decentralised participants, is built on the bedrock principle of reliability & scalability. Our distributed network ensures we are global - not reliant on a single data centre or network. Downtime simply doesn't make sense in this world.
02
Unrivaled inference speeds
Boasting a sub-second 'time to first byte' for LLM's, Subnet 19 delivers the capacity to process millions of tokens every second with remarkable speed. SOTA images come in seconds, pre-safety checked for NSFW content. There is no other place in the world you can get on demand, dedicated, decentralised inference to models this good.
03
Smart
Routing
Subnet 19 utilizes a decentralized network architecture to ensure high availability and uninterrupted access to the network. If one node in the subnet is unavailable to handle a user's request, the query is automatically routed to the next available node. This redundancy allows the network to continue operating smoothly even if individual nodes go offline. You will get your response, and you will get it fast!
04
Multiple Decentralised Entrypoints
Subnet 19 provides access to its diverse range of models through a decentralized pool of 20 validator nodes. This architecture eliminates any single point of failure, ensuring users can always connect to an available miner node to handle their requests.
ExpandingthecapabilityandscalabilityofdecentralisedAImodels.
Models available, directly through Subnet 19
Subnet 19 is a canvas for your creativity, offering boundless opportunities to generate, edit, or recreate images, chat with 3 Llamas, or even transform into a digital avatar - your imagination is the only limit
,,
Tokens per second available through Subnet 19
Backed by DSIS research, we ensure miners are incentivised to dedicate bandwidth to all validators proportional to their stake - maximising throughput.
GetstartedonchainwithTauvisiontoday.
