AI TrustMark

Model hosting / inference companies, products & suppliers

Model hosting and inference platforms run someone else's or your models as an API with scaling, isolation and observability. Buyers are purchasing infrastructure, not a chatbot. Tenancy, region and which models are actually served are the entity facts that matter.

What is Model hosting / inference?

Model hosting and inference platforms run someone else's or your models as an API with scaling, isolation and observability. Buyers are purchasing infrastructure, not a chatbot. Tenancy, region and which models are actually served are the entity facts that matter.

What problems does it solve?

Teams cannot serve models with isolation, regional control and cost limits.

Typical business use cases

  • Production inference
  • Private model APIs
  • Burst GPU

Important capabilities

  • Autoscaling
  • VPC
  • Routing
  • Observability

What buyers should evaluate

  • Tenancy
  • Regions
  • Supported runtimes
  • Keys

Risks and governance considerations

Cross-tenant prompt leakage.

Procurement checklist

  • Isolation evidence
  • Regions
  • KMS
  • SLAs

Relevant AI Trustmark assurance

Hosting quality is operational evidence, not a model-quality score.

Methodology · How verification works

Products and providers

Open a product profile for its catalogue details, or use the provider link for the company profile. Claimed listings appear first, then by name. Payment never changes the order.

  • Amazon Bedrock

    By Amazon Web Services

    Amazon Bedrock is AWS's fully managed service for building generative AI applications and agents. It offers access to foundation models from Amazon and other model providers throug

  • Baseten Model APIs

    By Baseten

    Baseten Model APIs provide on-demand access to leading open-source models, such as DeepSeek, running on the Baseten Inference Stack. The APIs are OpenAI-compatible, support functio

  • Cerebras Inference

    By Cerebras

    Cerebras powers the world's fastest AI inference on the biggest wafer chip. Cerebras CS-4 delivers up to 30x faster inference than GPUs. Cerebras Inference is Cerebras's model

  • Cloudflare Workers AI

    By Cloudflare

    Cloudflare Workers AI is Cloudflare's platform, from Cloudflare, Inc. Published at developers.cloudflare.com/workers-ai. Directory filing: Model hosting / inference. Vendor legal n

  • DeepInfra

    By DeepInfra

    DeepInfra offers cost-effective, scalable, production-ready ML model hosting and inference infrastructure. Public marketing also emphasises cost-focused serverless inference. It is

  • fal Serverless

    By fal

    fal Serverless is fal's platform, from Features & Labels, Inc. Published at fal.ai/serverless. Directory filing: Model hosting / inference. Vendor legal name on this product record

  • Fireworks Serverless Inference

    By Fireworks AI

    Fireworks’ state of the art training and inference platform take you beyond the frontier, transforming open models into your specialized intelligence. Fireworks Serverless Inferenc

  • GroqCloud

    By Groq

    Groq is the premier neocloud for fast inference. One fully integrated platform for infrastructure, inference, and control. Millions of developers run trillions of tokens on Groq ev

  • Hugging Face Inference Endpoints

    By Hugging Face

    Hugging Face operates a model hub, inference tooling and collaboration products for machine-learning teams. Public pages cover hosting, inference endpoints and open-model distribut

  • Hyperbolic

    By Hyperbolic

    Hyperbolic provides GPU cloud and inference services aimed at affordable access to AI compute and open models. Public marketing also emphasises price-competitive GPU/inference posi

  • Microsoft Foundry Models

    By Microsoft

    Microsoft Foundry Models is the model catalogue inside Microsoft Foundry for discovering, comparing and deploying AI models. It lists more than 10,000 models, from foundation and r

  • Novita AI

    By Novita AI

    Novita AI markets GPU cloud and model APIs for generative AI workloads. Public marketing also emphasises combined GPU cloud + model API offering. It is presented for Builders needi

  • NVIDIA NIM / NeMo

    By NVIDIA

    NVIDIA NIM and NeMo provide inference microservices and tooling to build, customise, and deploy AI models on NVIDIA infrastructure. Public marketing also emphasises gPU ecosystem s

  • OpenRouter

    By OpenRouter

    The unified interface for every model. Find the best models & prices for your prompts. OpenRouter is OpenRouter's platform, from OpenRouter, Inc. Published at openrouter.ai. Direct

  • Replicate API

    By Replicate

    Run open-source machine learning models with a cloud API. Replicate API is Replicate's model API, from Replicate, LLC. Published at replicate.com/docs/reference/http. Directory fil

  • Together Inference

    By Together AI

    Build what's next on the AI Native Cloud. Full-stack AI platform for inference, fine-tuning, and GPU clusters — powered by cutting-edge research. Together Inference is Togethe

Products

Claimed products appear first. Ranking packs and payment do not change this list.

Also used in this category

These products have a different primary category so they do not compete for the same ranking queries. They are listed here because buyers still encounter them in this job.

Related categories

Relevant procurement and assurance guides

Frequently asked questions

Is this an LLM vendor?

Not necessarily. Many hosts serve third-party or customer weights.

What is Model hosting / inference?

Model hosting and inference platforms run someone else's or your models as an API with scaling, isolation and observability. Buyers are purchasing infrastructure, not a chatbot. Tenancy, region and which models are actually served are the entity facts that matter.

What model or provider changes should a buyer insist on being told about?

Material change usually includes a new model family, new region, new subprocessor, new write-capable tool, or a change that affects logging, privacy or human oversight. Those changes should trigger evidence refresh rather than a silent release.

How should buyers verify where AI customer data is processed?

Ask for the named processing locations, cloud regions and any subprocessors that see prompts, files or outputs. A directory listing is not evidence of residency. Independent assessment records the locations that were in scope on the assessment date.

Does a TrustMark on one product cover the rest of the company?

No. Independent assessment is scoped to the named organisation and, where relevant, the named product. Category pages list suppliers as a topic label. They do not imply that every listed company has been assessed.

What incident-handling evidence is useful for AI suppliers?

Buyers should see how AI-specific failures are detected, contained and notified — including unsafe outputs, data leakage and unauthorised agent actions. An incident policy that never mentions models, prompts or tools is incomplete for this class of product.

What security testing evidence should buyers request for an AI product?

Ask what was tested, against which version, whether prompt-injection, data-exfiltration and tenant isolation were in scope, and where failed prompts were stored. A generic ISO certificate or a vendor scanner screenshot is not by itself an AI TrustMark assessment.

How should buyers check AI subprocessors?

Ask for the named subprocessors that process prompts, files, embeddings or logs, their locations, and whether a change is notified. Related-organisation hosting should be disclosed. A privacy policy URL is not a current subprocessor list.