Together is an AI cloud platform that provides infrastructure and services for training, fine-tuning, and deploying large language models and other generative AI systems. Founded in 2022, the company offers access to open-source models through managed APIs and high-performance GPU clusters, positioning itself as an alternative to proprietary AI providers like OpenAI and Anthropic. Together emphasizes open-source collaboration and aims to make advanced Artificial intelligence capabilities accessible to developers and enterprises.
The platform supports a range of workloads, from Machine learning experimentation to production-scale inference. It provides both serverless and dedicated compute options, with support for popular frameworks such as Transformer (architecture) architectures and Deep learning libraries. Together's infrastructure is designed to handle the demands of Large language model training and serving, offering low-latency responses and scalable resources.
History and Founding
Together was founded in 2022 by a team of researchers and engineers with backgrounds in Artificial intelligence and distributed systems. The company emerged from the growing need for accessible, cost-effective AI infrastructure, particularly for open-source models. Its founders include Vipul Ved Prakash, Ce Zhang, Chris Re, and Percy Liang, who brought experience from academia and industry. The company quickly gained attention for its focus on open-source AI, distinguishing itself from closed platforms.
In its early years, Together raised significant funding from venture capital firms, enabling rapid expansion of its GPU clusters and service offerings. By 2023, it had become a notable player in the AI cloud market, competing with larger providers like Amazon Web Services and Google Cloud. The company also launched partnerships with model developers and research institutions, further cementing its role in the AI ecosystem.
Services and Products
Together offers a suite of services centered around generative AI. Its primary products include:
- Inference API: A managed service for running open-source models, such as Llama and Mistral, with simple REST endpoints. This allows developers to integrate AI capabilities without managing underlying infrastructure.
- Training and Fine-Tuning: Tools for customizing models on proprietary datasets, using distributed training across multiple GPUs. Together provides pre-configured environments and optimization techniques to reduce costs.
- GPU Clusters: Dedicated compute resources for large-scale training jobs, with options for various GPU types and network configurations. These clusters are designed for high throughput and low latency.
- Together Enterprise: A tier for organizations requiring enhanced security, support, and compliance features, including private networking and dedicated instances.
The platform supports Generative AI applications such as chatbots, code generation, and content creation. It also offers compatibility with popular libraries like Hugging Face Transformers, simplifying migration for existing users.
Technology and Infrastructure
Together's infrastructure is built on a distributed computing architecture that leverages high-performance GPUs, including AMD and NVIDIA hardware. The company operates data centers in multiple regions, ensuring redundancy and low-latency access for global users. Its software stack includes custom schedulers and networking optimizations to maximize utilization and minimize bottlenecks during training.
A key differentiator is Together's emphasis on open-source models. Unlike OpenAI or Anthropic, which primarily offer proprietary systems, Together provides access to community-developed models, allowing users to inspect, modify, and deploy them freely. This aligns with the broader open-source AI movement, which advocates for transparency and collaboration in Machine learning research.
The company also invests in research to improve efficiency, such as novel attention mechanisms and quantization techniques. These efforts aim to reduce the computational cost of Large language model inference, making AI more sustainable and accessible.
Market Position and Competition
The AI cloud market is highly competitive, with major players including Amazon Web Services, Google Cloud, and Microsoft Azure offering AI-specific services. Together differentiates itself through its focus on open-source models and developer-friendly APIs. It also competes with specialized AI cloud providers like Coreweave and Cerebras, which offer high-performance compute for AI workloads.
Together's pricing model is usage-based, with per-token or per-hour rates, making it attractive for startups and research groups. The company has reported strong growth, with customers spanning industries such as finance, healthcare, and media. As of 2024, Together continues to expand its offerings, including support for multimodal models and edge deployment.
Community and Open Source
Together actively contributes to the open-source AI community by releasing tools, benchmarks, and model weights. It sponsors research projects and hosts events to foster collaboration among developers and researchers. The company's commitment to openness is reflected in its support for initiatives like the Open Source Initiative and partnerships with academic institutions.
By providing accessible infrastructure, Together aims to democratize AI, enabling smaller organizations to compete with tech giants. This mission resonates with many in the Artificial intelligence field who advocate for broader participation in AI development.
Future Directions
Looking ahead, Together plans to enhance its platform with more advanced features, such as automated model optimization and multi-cloud support. It is also exploring ways to reduce energy consumption through more efficient hardware and algorithms. As Generative AI continues to evolve, Together is positioned to play a significant role in shaping how these technologies are deployed and accessed.
The company faces challenges, including competition from well-funded rivals and the rapid pace of AI innovation. However, its focus on open-source principles and developer experience provides a strong foundation for sustained growth.