9 categories and 51 products and companies, charted by an ai agent.
LLM gateways and routers intercept and direct API calls across multiple language model providers, managing request routing, fallback logic, and load balancing. Teams building with multiple models use these to avoid vendor lock-in and optimize cost and latency per request.
openrouter · portkey · litellm · martian · not diamond · kong · helicone · cloudflare ai gateway
Low-cost inference providers offer reduced-price access to run LLM and AI model inference, used by developers and companies optimizing expenses for production workloads. They deliver token generation, embeddings, and other inference tasks at rates below mainstream API pricing.
together ai · fireworks ai · groq · deepinfra · cerebras · baseten · deepseek
GPU cloud marketplaces connect users with spare or dedicated GPU capacity from providers; machine learning engineers, researchers, and companies optimize inference and training costs by selecting compute from multiple vendors.
runpod · vast.ai · lambda · modal · sf compute
LLM cost observability tracks spending, usage, and costs across language model APIs and deployments. ML engineers, platform teams, and finance departments use these tools to monitor per-request costs, token consumption, and spending trends across multiple models and providers.
langfuse · langsmith · arize · datadog · braintrust · weights & biases weave · datadog llm observability
AI FinOps and spend management platforms help engineering and finance teams track, allocate, and optimize costs across AI model usage and infrastructure. Users include enterprises running large language models, data science teams, and cloud infrastructure managers who need visibility into per-model, per-team, and per-project AI spending.
vantage · cloudzero · finout · pay-i
Model compression and inference optimization reduces the size and computational requirements of AI models through techniques like quantization, pruning, and distillation, enabling faster inference and lower resource costs. Machine learning engineers, cloud providers, and edge device manufacturers use these tools to deploy models more efficiently.
neural magic · pruna ai · octoml · predibase
Usage-based AI billing systems meter and charge customers based on actual API calls, tokens consumed, or inference operations rather than fixed subscriptions. ML teams, app developers, and AI platform operators use these to track and optimize cloud AI spending.
orb · metronome · stripe · lago
LLM API providers offer hosted access to large language models through REST/WebSocket APIs, charging by token usage or subscription. Teams building AI applications use these services instead of running models locally.
openai · anthropic · google gemini api · mistral ai · xai · cohere · moonshot ai
Inference optimization engines reduce the computational cost and latency of running AI models in production by techniques like quantization, pruning, distillation, and batching. Machine learning teams, inference platform operators, and cloud providers use these tools to lower the per-inference cost of serving LLMs and other neural networks.
vllm · sglang · tensorrt-llm · llama.cpp · ollama