GenAI LLM Platforms

Vendors that can provide enterprise Generative AI (GenAI) platforms, secure Model API access (“Model API”) to the Vendors’ Frontier Models, and related services, including implementation, integration, adoption, AI literacy, and role-based upskilling, for use by authorized users and authorized State applications.
- GenAI Offering(s), including the user-facing platform, administrative capabilities, secure Model API access, available implementation and integration services, AI literacy and role-based upskilling services, main features, common public-sector use cases, and how the offering can improve State Government agency operations.
- GenAI platform supports the following and whether it is generally available, in
preview, in development, or planned; whether it is native or relies on third-party products or custom
development; and any material licensing, regional, capacity, or usage limitations.
• Supports natural-language interactions for summarization, content generation, rewriting, translation, classification, information extraction, conversational question answering, reasoning, and analysis.
• Ingests, searches, summarizes, compares, and analyzes individual documents and collections of documents, including native and scanned documents, presentations, spreadsheets, structured data, images, audio, and video. Identify applicable file, page, context, and collection limits.
• Maintains context across multi-turn conversations, including prior messages, instructions, attachments, and generated artifacts, and allows users to resume, branch, search, share, or export conversations.
• Provides optional persistent memory for users, projects, workspaces, or agents. Describe how memory is scoped, inspected, corrected, disabled, retained, exported, and deleted.
• Supports brainstorming, planning, and the creation or editing of documents, presentations, spreadsheets, reports, forms, diagrams, charts, images, and other structured outputs.
• Processes and generates combinations of text, images, audio, video, tables, charts, diagrams, forms, and structured data, including optical character recognition, document-layout understanding, chart interpretation, and visual question answering.
• Supports speech-to-text, text-to-speech, translation, captioning, speaker identification, and real-time voice interaction. Identify supported languages, latency, transcript-editing, and audio-retention capabilities.
• Analyzes structured and unstructured data, performs calculations, generates formulas and code, creates visualizations, and produces reproducible analytical outputs. Describe controls and isolation for any code-execution environment, whether cloud or desktop.
• Grounds responses in supplied or retrieved sources and provides verifiable citations to the applicable page, section, cell, row, timestamp, or record where available. Describe how the platform identifies conflicting, outdated, missing, or insufficient evidence and communicates uncertainty.
• Retrieves and synthesizes current information from public, licensed, and Stateapproved sources. Describe source controls, retrieval dates, citations, and the ability to restrict or disable external research.
• Supports multiple Vendor model families and versions, including model version selection, task-based routing, fallback, version pinning, evaluation, migration, and the ability to enable or disable individual models.