Baseten

Provides an inference platform to serve and scale open-source, custom, and fine-tuned AI models for production workloads, addressing performance, latency, cold starts, and reliability at massive scale. Serves developers, data teams, and IT teams in B2B companies building generative AI applications across modalities such as LLMs, transcription, image generation, and text-to-speech. Delivered as a managed cloud platform with single-tenant, self-hosted VPC, and hybrid deployment options across clouds and regions with 99.99% uptime.

Key features

• Serve open-source and custom AI models
• Dedicated inference for high-scale workloads
• Pre-optimized Model APIs for instant evaluation
• Train models with Loops SDK and RL
• Deploy training outputs to production inference
• Custom kernels and decoding techniques
• Advanced caching and quantization
• Cross-cloud high availability
• Fast cold starts with 99.99% uptime
• Single-tenant and self-hosted VPC deployment
• Hybrid deployment with flex capacity
• Support for LLMs and compound AI
• Optimized transcription and diarization
• Real-time text-to-speech with streaming
• Image generation for custom and ComfyUI models
• Forward deployed engineering support