At Baseten we provide all the infrastructure you need to deploy and serve ML models performantly, scalably, and cost-efficiently. 

Get started in minutes, and avoid getting tangled in complex deployment processes. You can deploy best-in-class open-source models and take advantage of optimized serving for your own models.

We also utilize horizontally scalable services that take you from prototype to production, with light-speed inference on infra that autoscales with your traffic.

Best in class doesn't mean breaking the bank. Run your models on the best infrastructure without running up costs by taking advantage of our scaled-to-zero feature.

AI Fund

Caffeinated Capital

DJ Patil

Lachy Groom

Mustafa Suleyman

South Park Commons

Cristina Cordova

Dev Ittycheria

Jay Simons

Baseten

Explore the capabilities of Playground v2 and Stable Diffusion XL 1.0 in our comprehensive comparison for text-to-image generation.

This analysis offers insights into each model's strengths, from creative potential to processing speed, guiding enthusiasts on the best fit for their projects. A must-read for those passionate about advancing their digital artistry.

#AI #DigitalArt #TechComparison

💡 Super excited and proud with the work that our team pulled together over the past weekend as part of Penn's inaugural Generative AI Hackathon - so much so that I wanted to share it here!

🛠 24 hours of tinkering, hacking, experimenting and building with Emma (Yidi) Xu, Yimeng Li, and Tanika Mehra culminated in the creation of StoryG - an AI-driven dynamic story writer for children that can write new or adapt existing fairytales and stories, customized by features such as age, interests, region/cultural identity, interactivity, duration, genre - and even specific morals or learning concepts that a parent/guardian wants emphasized. 

📖 Common fairytales lack the representativeness to be relevant in today's diverse world, and yet they are still effective tools to help children learn about complex moral issues. With StoryG, parents and guardians can create compelling, relevant, and representative stories that can speak to a child's unique, individualized context - in mere seconds and in multiple modalities! Our mission is to use the power of telling stories and AI to make every child the hero of their own journey.

🏫 My motivation for this idea came from my issue with where generative AI is going today as an industry. This technology has the power to truly change the way we interact with information, and if used effectively, can also democratize access to information in unique and innovative ways - but it will not do so as chatbots powering enterprise search, as LinkedIn post-writing copilots, or as search engine assistants. It will do so by augmenting and innovating our foundational industries such as healthcare, education, and early childhood development. This is where generative AI can be truly revolutionary to society.

🏆 Thank you to my team once again for the incredible effort. We finished as finalists in all of the Social Impact, Best GPT wrapper, Best rookie project, and Best overall project categories!

💻 And thank you to Contrary, Pinecone, Runway, Nomic AI, Weaviate, Z Fellows, Bronco AI, Baseten, CommandBar, Macro for putting this together and giving me an excuse to nerd out and create something really cool with your tools and to Venture Lab for hosting!

PS: see the last slide for early user feedback!

#generativeai #genai #ai #artificialntelligence #hackathon #gpt #gpt4 #openai #venturelab #technology #product #productinnovation #responsibleai

Boost your AI's speed and efficiency with Baseten's breakthrough in Mixtral inference!

By leveraging TensorRT-LLM and quantization, we've achieved significant performance milestones on a single A100 GPU, including a time to first token of 127ms and 81 perceived tokens per second. Discover how to cut inference costs in half while maintaining quality.

#AI #MachineLearning #Quantization

Master the art of benchmarking AI models like Stable Diffusion XL.

Our guide explores key metrics such as latency and throughput, vital for enhancing AI efficiency. Perfect for developers seeking performance optimization.

#AI #Benchmarking #StableDiffusionXL

Quantizing ML models: a game-changer for AI performance and cost efficiency.

Philip and Abu unveil how reducing the precision of model weights enhances inference speed and reduces hardware demands. A pivotal read for anyone looking to optimize AI deployments.

#MachineLearning #AI #Quantization

https://lnkd.in/edDrc7mc

What is the best open source large language model?

The answer, as usual: it depends.

Here are 6 recommendations organized by budget and use case:

https://lnkd.in/gFx7kvyM

Launching today 🎉

Double your throughput or halve your latency for Mistral AI, Stability AI + others? 

Do both at ~20% lower cost with NVIDIA H100s on Baseten.

Using NVIDIA AI TensorRT, Baseten has unlocked the full potential of the H100 architecture for lightning fast inference. With TensorRT, customers get 2-3x higher inference throughput at ~20-40% lower cost than an equivalent A100.

https://lnkd.in/dBqJfeUY

Happy feature Friday! 

Excited to introduce brand-new integrations with model providers and GPU clouds within HoneyHive 🔌

Get Day 1 access to any closed or open-source LLM in the Playground, with built-in version control, logging, and benchmarking on your own datasets 📈

Supported Model providers: OpenAI, Anthropic, Cohere, AI21 Labs, Mistral AI, Aleph Alpha, Perplexity and more

Supported GPU Clouds: Anyscale, Together AI, Hugging Face, Baseten, Replicate, Amazon Web Services (AWS), Google, Microsoft and more

Learn about how our collaboration with #AWS accelerates innovation for Generative AI startups. Join industry experts from #NVIDIAInception members Baseten and Union.ai on AWS On Air.

Watch Now > https://nvda.ws/3GAXpr5

Type in any destination and teleport there in < 5s.

It's like Google Street View for your imagination.

David Song and I spent the last week messing around with latent consistency models (LCMs) and Stable Diffusion to generate panoramic images very quickly based on a prompt. It's a ton of fun on mobile. 

🚀 Try it here: https://latentverseai.com
👨💻 Code: https://lnkd.in/gnvvF-Nb

Everything is hosted on Baseten for super fast model inference 🏎️ 💨

Let us know what you think!

Baseten is a serverless backend that offers a fast way to build machine-learning-powered applications with auto-scaling, GPU access, CRON jobs, and serverless functions. Data science and machine learning teams can use Baseten to build applications without prior knowledge of backend, frontend, or MLOps, allowing for faster, smarter decision-making at their organizations.

CO:BASETEN

Baseten

Filings Not Available