Helicone
Open-source LLM observability platform for logging, monitoring and debugging LLM API calls -- now in maintenance mode after Helicone's acquisition by Mintlify; new customers are directed to migrate rather than adopt it.
What it does
- Evaluation and observability
Used for
- Guardrails and safety
Features 15
- SessionsGroups related LLM calls, vector queries and tool calls so an entire agent flow can be traced in one view.
- Custom PropertiesTag LLM requests with metadata for filtering, per-user or per-feature cost analysis and performance tracking.
- Cost Tracking & OptimizationMonitors LLM spending across providers and offers optimization tools for managing the AI budget.
- DatasetsCurate and export LLM request and response data for fine-tuning, evaluation and analysis.
- AlertsNotifies by email or Slack when LLM applications hit error-rate or cost thresholds.
- WebhooksSends request and response data to your endpoint through a unified request_response_url payload.
- ReportsSends automated weekly reports on how your LLM application is performing.
- HQL (Helicone Query Language)Lets you query your Helicone analytics data directly with SQL.
- Eval ScoresLets you report evaluation results from any evaluation framework into Helicone.
- User FeedbackCollects positive or negative ratings on LLM responses.
- LLM CachingStores complete LLM responses on Cloudflare's edge network for reuse.
- LLM Rate LimitsLets you set tailored rate limits on calls to model provider APIs.
- Prompt ManagementCompose and iterate on prompts, then deploy them in LLM calls through the AI Gateway.
- LLM SecurityBuilt-in security checks for LLM requests, powered by Meta's security models.
- ModerationsChecks whether a user message is potentially harmful.
How it compares
-
Braintrust
As teams evaluate Helicone alternatives, understanding how different tools handle observability becomes critical to long-term success.
Official documentation · 13 Oct 2026 -
LangSmith
Helicone is also more cost-effective than LangSmith as it operates on a volumetric pricing model.
Official documentation · 13 Oct 2026
Sources
- Pricing
-
Official documentation · 31 Aug 2026
Pricing page lists Hobby (free, 10,000 requests/1GB storage), Pro ($79/month), Team ($799/month), Enterprise (contact-sales). No mention of maintenance mode or wind-down on this specific page.
- Used for
-
Official documentation · 24 Oct 2026
'Enable robust security measures in your LLM applications to protect against prompt injections, detect anomalies, and prevent data exfiltration.' (https://docs.helicone.ai/features/advanced-usage/llm-security.md)
- Sold within
-
Helicone's pricing page offers "Hobby" ("Free", "10,000 free requests", "Get started for free") and paid plans from "$79". The site carries "Helicone Joins Mintlify"; the announcement says "Helicone's services will remain live for the foreseeable future in maintenance mode."
- Deployment
-
Official documentation · 13 Oct 2026
Page: 'Deploy your own instance of Helicone using your preferred method' / 'Sign up for free and complete the onboarding flow'.
- Interfaces
-
Official documentation · 14 Oct 2026
Page: 'Navigate to Settings → Reports in your Helicone dashboard' / 'Helicone's AI Gateway is an OpenAI-compatible, unified API with access to 100+ models, including OpenAI, Anthropic, Vertex, Groq, and more' / 'Log LLM responses from any external tools used in your LLM applications using Helicone's Logger SDK' / 'Query your Helicone observability data directly from MCP-compatible AI assistants using the Helicone MCP server'.
- Modalities
-
Official documentation · 14 Oct 2026
Page: 'Messages: map your trigger fields (e.g., prompt text) into the user message' / 'Use the Helicone app on Zapier to generate chat completions from any supported model through the Helicone AI Gateway' / 'Helicone's AI Gateway supports image generation through models with native image output capabilities'.
- Intended for
-
"open-source LLM observability and monitoring platform for developers"
Also from Helicone
-
Helicone AI Gateway API service
OpenAI-compatible unified API routing to 100+ models (OpenAI, Anthropic, Vertex, Groq and more) through Helicone's platform.
Open Helicone in the directory
Something wrong here? Send a correction — quote this product id: helicone-platform.