- Home
- Model Monitoring Software
- Memoturn
Memoturn
What it is
Memoturn is an open-source platform for engineering applications built with large language models. It brings together tracing, cost and latency metrics, evaluations, prompt management, monitors, and datasets. Its waterfall timeline displays traces, spans, generations, and scores, with session views and per-user rollups. Teams can run offline dataset experiments, sample production traces with online evaluators, and send examples to human review queues for scoring. A versioned prompt registry supports production, staging, and custom channels, plus A/B experiments that divide live traffic between prompt versions. Memoturn accepts OpenTelemetry data using OTLP/JSON and GenAI semantic conventions, and provides TypeScript, Python, and Go SDKs, a REST API, and an OpenTelemetry receiver. Its integration guide lists wrappers for model providers, frameworks, and vector stores, though some first-party wrappers are limited to TypeScript or Python. Deployment options include local operation, Docker Compose, and Helm on Kubernetes. Telemetry storage can use Apache Doris or PostgreSQL, with PostgreSQL described for smaller installations. The Apache-2.0 core includes access controls, audit logs, PII masking, retention policies, exports, API-key management, and rate limits.
For and against
For
- Open-source core licensed under Apache-2.0.
- Tracing views include sessions and per-user rollups.
- Supports offline experiments and online evaluators.
- Prompt registry includes versioning and live-traffic A/B experiments.
- Can run locally, with Docker Compose, or on Kubernetes.
Against
- Some first-party integration wrappers are TypeScript-only or Python-only.
- PostgreSQL is described for smaller installations.
- No paid tier or usage limits are stated on the opened pages.
The verdict
Memoturn brings LLM tracing, evaluations, prompt controls, and operational metrics into one self-hostable platform. Check its language-specific integration coverage and deployment requirements against your stack.
Who it suits
Memoturn suits teams building LLM applications that need to inspect production traffic, monitor spending, evaluate quality, and manage prompts. Its self-hosting options are relevant to operators who want telemetry to stay on infrastructure they own.
Plans and prices
Memoturn (Apache-2.0 self-hosted core)
No paid tier or usage limits stated on the opened pages
Read 7 Oct 2026 · memoturn.com
On the record
- Agent IDE support
- Its MCP server exposes prompts, datasets, and review queues to agent IDEs such as Claude Code and Cursor, locally or over Streamable HTTP.memoturn.com · read 4 Oct 2026
- Automations
- Alert rules and cost budgets are evaluated every minute, and event rules can send webhooks or Slack messages when scores drop or spend spikes.memoturn.com · read 4 Oct 2026
- Data location
- Memoturn describes itself as self-hostable and says telemetry stays on infrastructure the operator owns.memoturn.com · read 7 Oct 2026
- Deployment
- The maker documents local setup, Docker Compose deployment, and Helm on Kubernetes.memoturn.com · read 7 Oct 2026
- Evaluations
- It supports offline dataset experiments, online evaluation of sampled production traces, and human review queues.memoturn.com · read 7 Oct 2026
- Integrations
- The maker lists OpenTelemetry, OpenAI, Anthropic, LangChain, LiteLLM, and MCP, and its integration guide documents additional model providers, frameworks, and vector stores.docs.memoturn.com · read 7 Oct 2026
- Intended users
- The maker positions Memoturn for teams building LLM applications who need to debug production traffic, track spend, catch quality regressions, and manage prompts.memoturn.com · read 7 Oct 2026
- License
- The project is licensed under Apache-2.0.memoturn.com · read 4 Oct 2026
- Metrics and alerts
- Dashboards report cost, tokens, and latency, while alert rules and cost budgets can trigger Slack messages or webhooks.memoturn.com · read 7 Oct 2026
- Notable integration limit
- The integration guide marks first-party wrappers for some integrations as TypeScript-only or Python-only, while Go coverage includes OTel export helpers and manual tracing.docs.memoturn.com · read 7 Oct 2026
- Product
- Memoturn is an open-source AI engineering platform for LLM tracing, evaluation, metrics, prompt management, and datasets.memoturn.com · read 7 Oct 2026
- Prompt management
- Its versioned prompt registry supports deployment channels and A/B experiments that compare versions by score.memoturn.com · read 7 Oct 2026
- SDKs
- The product lists TypeScript, Python, and Go SDKs.memoturn.com · read 4 Oct 2026
- SDKs and API
- The maker lists TypeScript, Python, and Go SDKs, plus an OpenTelemetry receiver and a REST API.memoturn.com · read 7 Oct 2026
- Security controls
- The Apache-2.0 core includes OIDC and SAML SSO, organization roles, audit logs, PII masking at ingest, retention policies, and scheduled NDJSON exports.memoturn.com · read 7 Oct 2026
- Self-hosting
- Memoturn says it is self-hostable and that telemetry stays on infrastructure the user owns.memoturn.com · read 4 Oct 2026
- Storage options
- The deployment guide supports Apache Doris or PostgreSQL as the telemetry engine, with PostgreSQL described for smaller installs.docs.memoturn.com · read 7 Oct 2026
- Telemetry
- Memoturn accepts OpenTelemetry data and supports OTLP/JSON with GenAI semantic conventions.memoturn.com · read 4 Oct 2026
