Helicone

Researched 2026-06-28. Primary category: llm-observability.

One-liner — Open-source, proxy-based LLM observability and AI gateway — one line of code to monitor LLM usage; acquired by Mintlify and now in maintenance mode.

What it does — Helicone sits as a proxy between your app and LLM providers: change one line (point your base URL at Helicone) and it logs every request/response, cost, latency, and usage, with caching, rate-limiting, and a unified gateway to 100+ models via an OpenAI-compatible endpoint. The low-friction proxy model was its signature. It processed ~14.2 trillion tokens across ~16,000 organizations.

Where it sits in the stack — The llm-observability layer; its gateway features also overlap ai-gateway. As a proxy it sits on the model-traffic path, but it is an observability/routing layer, not a security firewall — it does not by itself screen prompts or block sensitive-data leakage. Adjacent to litellm, portkey, openrouter on the gateway side.

Deployment & architecture — Inline proxy (the default, one-line integration), plus async logging SDK option; self-hostable OSS or Helicone Cloud SaaS. Integrates with OpenAI, Anthropic, Azure, and other providers at the API layer.

Positioning & differentiators — Best known for the proxy-based, one-line-change integration — the lowest-friction onboarding in the category and a combined observability + gateway pitch. Contrast with SDK/OTel-based langfuse, arize-phoenix, langsmith (more eval depth, deeper instrumentation) and eval-first braintrust. The proxy model adds a network hop on the critical path, a tradeoff for the convenience.

Ownership, funding & M&A — Acquired by Mintlify, announced 2026-03-03 (confirmed via both companies’ blogs). Founders Justin Torre and Cole Gottdank joined Mintlify in San Francisco; the Helicone product remains live but in maintenance mode (security patches, new-model support, bug fixes only — active feature development ended). Founded 2023 (YC W23), HQ San Francisco; funding was early-stage/small. Confidence: high.

CTO / hedge-fund lens — The maintenance-mode status makes Helicone a weak choice for a new Day-1 deployment — you would be adopting a frozen product. Of interest mainly to teams already on it (self-hosted OSS can keep running) or studying the proxy pattern. For new builds prefer an actively developed alternative. The inline-proxy model also means LLM traffic (potentially MNPI) flows through Helicone unless self-hosted — self-host if you adopt it at all.

Competitors / alternatives — langfuse, langsmith, arize-phoenix, braintrust, datadog, comet, litellm, portkey.

Open questions / to verify

• Long-term fate of Helicone Cloud SaaS under Mintlify (how long it stays live).
• Whether the OSS repo continues to receive community contributions.

Sources

• Helicone is joining Mintlify (Helicone blog) — fetched 2026-06-28 — supports: acquisition by Mintlify 2026-03-03, maintenance mode, proxy model, scale, founders; confidence: high.
• Mintlify acquires Helicone (Mintlify blog) — fetched 2026-06-28 — supports: acquisition confirmation/rationale; confidence: high.

History

• [2026-06-28] Stub created from seed registry.
• [2026-06-28] Researched; established Helicone acquired by Mintlify (2026-03-03, confirmed), product now maintenance-mode — changed ownership independent→acquired (confidence low→high), set hedge_fund_fit low for new deployments; filled founding/HQ/proxy deployment. No seed M&A flag existed but acquisition found and recorded.