We’re excited to work with Together AI to bring open models to Daily Bots. Developers can build voice-to-voice AI and real-time multimodal agents with all three size of Llama 3.1.

LlamaRank is SOTA reranking model from Salesforce AI Research. Now available (exclusively) on Together AI API. For RAG pipelines and other ranking tasks. Also works with semi-structured data like JSON.

Excited to finally bring this product to market! The kernel team at Together AI kernel team led by Tri Dao has optimized a plethora of operators, some fairly fundamental ones, and made them available as kernels that can be registered in your training loop or inference

Vipul Ved Prakash Artificial Analysis NVIDIA 100 TPS! Amazing!!

Should we host Reflection 70B?

The future of LLM inference is quite likely to move in this direction: distilling transformer models into Mamba-Transformer hybrids with minimal quality loss but with significant performance and throughput benefits. The results with Llama-3 8B are very strong!

“Human societies have absorbed new disruptive technologies over millennia and remade themselves to thrive with their aid—and artificial intelligence will be no exception.” My op-ed in TIME on the long road to peak AI and why AGI talk and regulatory interventions are premature.

There's a good argument Llama will be the default pre-trained model given Meta's compute footprint, access to data, research bench, and open-weights approach, and a lot of practical AI R&D might happen at startups and labs focused on post-training llamas.

The Together AI research and applied ml group will be at PyTorch conference. If you are interested in developing accelerated systems for inference and training, please come see at booth S24.

Llamacoder is wild.

1/ We're excited to launch Together Talks, a series of conversations with researchers and industry leaders on generative AI, brought to you by Together AI. 🎙️ In this series, you’ll hear from Vipul Ved Prakash, Percy Liang, Ce Zhang, Tri Dao, and Shuaiwen Leon Song from Together AI;

Together AI's inference and continuous performance optimization platform is now available in virtual private clouds.

Llama 405B now 30% cheaper on Together API.

Together API now provides _free_ access to Llama 3.2 11B Vision model. Since we founded Together AI, I've wanted to offer to a great foundation model for free, so creative developers everywhere can build AI enabled apps without cost. We are able to do this now. Thank you to

🚀 We've partnered with AI at Meta to offer Llama 3.2 11B Vision for FREE so developers can experiment with open-source multimodal AI at no cost. In addition to our free credits, for a limited time developers can get unlimited access to Llama 3.2 11B Vision via our

Vipul Ved Prakash Damn, this is pretty good!

Multimodal goes brr Together AI