The essentials: The Model Profiler aggregates model metadata from seven data sources into a single web interface with filtering, comparison, and availability maps to accelerate model selection for CTOs.
AWS releases the Model Profiler as an open-source tool to simplify the selection of foundational models in Amazon Bedrock. The solution centralizes metadata from over 100 models from various providers in a searchable interface.
The Model Profiler is a web application that centralizes access to more than 100 foundational models from providers such as Anthropic, OpenAI, Meta, Mistral AI, Cohere, and Amazon. Previously, decision-makers had to research model specifications, pricing information, regional availability, and context window size across multiple AWS console pages, documentation, and APIs. The tool aggregates this data from seven sources: five AWS APIs (ListFoundationModels, Price List, Service Quotas, ListInferenceProfiles, and Mantle) plus two public data sources (LiteLLM Model Database and AWS documentation).
For CTOs, the mapping of performance metrics is particularly relevant: tokens-per-minute (TPM) defines the throughput ceiling — approximately 1,000 tokens correspond to around 750 words of text. Requests-per-minute (RPM) independently limit the number of API calls. The tool displays these quotas per model and region and enables direct cost comparisons between on-demand, batch, and reserved tier pricing across 33 AWS regions. In addition, model status (active, legacy, end-of-life) is mapped from AWS documentation.
Data collection runs fully automated via AWS Step Functions with 17 Lambda functions. An agentic system powered by Amazon Bedrock identifies data gaps and applies self-healing corrective configurations. Through S3-based caching between Lambda functions, the number of API calls is reduced from approximately 480 to 29 per run (97% cache hit rate). The entire pipeline takes 8–12 minutes and runs daily to keep pricing and availability data current. The solution is available as open source and can be deployed in under five minutes.
Source: aws.amazon.com · Published July 1, 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrase and classification via Lumi News Pipeline v1.7.2.