Mnexium

Mnexium is an AI memory API for LLM apps and agents. It adds persistent memory and context across sessions so applications can remember facts, preferences, chat history, user profiles, records, and live context without building memory infrastructure separately. The site describes it as working with OpenAI, Anthropic, and Gemini, and as sitting between an app and the model to add memory and context to requests.

Its API includes options to learn facts from conversations, recall relevant memories, and prepend chat history. The product page also describes chat history preservation, agent memory that learns facts and recurring context over time, records for structured app data such as accounts, tickets, tasks, and deals, agent state for tracking tasks and workflow progress, integrations for external APIs and webhooks, and observability for tracing decisions, memory recall, tool usage, and API calls. Search by semantic similarity is shown in the docs, along with named methods for subject-based memory access. The examples are presented in cURL, Node.js, and Python, and the page shows a drop-in integration using the Mnexium SDK and a base URL at mnexium.com/api/v1.

Pricing is listed in tiers. A Free plan requires no signup and is described for building and testing AI apps with persistent memory. Paid plans include Builder at $29 per month, Growth at $149 per month, and an Enterprise plan with custom pricing, custom deployment models, priority SLA, and onboarding support. The page also says API access is available on the free plan. Mnexium is therefore an AI memory and context layer for applications and agents, delivered as an API and SDK-based service rather than as a standalone end-user app.