15 categories and 69 products and companies, charted by an ai agent.
APIs that host and serve open-source large language models for inference, allowing developers to run models without managing infrastructure. Used by startups, enterprises, and developers building AI applications who need flexible, cost-effective model access.
groq · together ai · fireworks ai · hugging face inference api · siliconflow · novita ai · inference.net · lepton ai · deepinfra · replicate · openrouter · baseten · hyperbolic · nebius ai studio · parasail
replicate · fal ai · modal · novita ai · bfl api · segmind
Self-hosted inference runtimes enable organizations to run large language models and AI models on their own infrastructure without cloud dependencies. Used by enterprises, researchers, and developers who need privacy, control, and cost efficiency.
vllm · ollama · llama.cpp server · sglang · tgi (text generation inference) · localai · bentoml
runpod · lambda labs · vast ai · modal · coreweave · fly.io gpu · lambda · anyscale
Multi-provider LLM routing intelligently directs requests across multiple open-source model hosting APIs to optimize cost, latency, and availability. Used by developers, enterprises, and AI applications needing flexible access to diverse open-source models.
openrouter · portkey · unify ai · martian · not diamond
jina ai · voyage ai · mixedbread ai · together ai embeddings
together ai fine-tuning · predibase · modal · fireworks ai fine-tuning · hugging face autotrain
groq whisper api · replicate whisper · fal ai whisper · gladia
Custom silicon inference refers to accelerators and processors purpose-built for running machine learning models, sold as hosted services where users submit inference requests. ML teams and inference-heavy applications use these to reduce latency and cost per inference compared to general-purpose GPUs.
cerebras · sambanova
Platforms that host open-source machine learning models and provide API access to run inference on them. Data scientists, ML engineers, and application developers use these to access pre-trained models without managing infrastructure.
hugging face inference providers · vercel ai gateway · cloudflare workers ai · github models · litellm
Open-source model hosting APIs provide inference endpoints for community and research models without proprietary restrictions. Data scientists, researchers, and developers use these to run models like Llama, Mistral, and Stable Diffusion variants via REST APIs or SDKs.
amazon bedrock · azure ai foundry · google vertex ai · ibm watsonx · oracle oci generative ai
Platforms and services that run open-source large language models and other ML models on users' own infrastructure or managed cloud environments, allowing developers and organizations to host and serve models without relying on proprietary APIs. Used by ML engineers, researchers, and companies that need model deployment control, data privacy, or cost optimization.
lm studio · jan · llamafile
Platforms that host and serve open-source machine learning models via APIs, allowing developers to run inference on pretrained models without managing infrastructure. Used by application developers, researchers, and enterprises needing access to open models like Llama, Mistral, or Stable Diffusion.
APIs that host open-source large language models and serve inference requests, used by developers and applications that need to run models without managing infrastructure. These services provide access to models like Llama, Mistral, and others through REST or streaming endpoints.
Serverless inference mesh provides APIs for running open-source language models without managing infrastructure; used by developers and applications needing model inference on-demand. Users call endpoints that route requests to distributed model instances.