# Helicone AI Gateway + MCP (slim)

> Open-source LLM observability that logs requests through an OpenAI-compatible gateway (100+ models, fallbacks, caching, rate limits) or async logging.

- Full: https://www.anchorterminal.com/tools/helicone.md (~6,850 tokens) · this version ~1,630 tokens · JSON https://www.anchorterminal.com/tools/helicone.json · canonical https://www.anchorterminal.com/tools/helicone
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · Updated: 2026-10-04

**D · 47.1/100 · rank #387 of 452 · #9 in Agent observability & evals · not agent-ready · confidence medium**


Assessment: One base-URL change gives logging, caching, fallbacks and rate limits for 100+ models. Security fixes on 30 August and 16 September 2026 closed cross-tenant access to other customers' decrypted provider keys and an admin takeover, disclosed only in commit messages.

## Facts

- Kind: HTTP API · vendor: Helicone (Mintlify) · category: Agent observability & evals · legal entity: Helicone, Inc. · provenance 82/100
- Endpoint: `https://ai-gateway.helicone.ai` (HTTP, stdio)
- Auth: API key · pricing: Freemium · x402: no · licence: Apache-2.0
- Probe metrics: not measured yet (probes haven't run)
- Free tier: Hobby, 10,000 requests a month, 1 GB storage, 1 seat, 7 days of retention, no card
- Rate limits: Ingestion 10 logs a minute on Hobby, 1,000 on Pro, 15,000 on Team, 30,000 on Enterprise. API 10 calls a minute on Pro, 60 on Team, 1,000 on Enterprise (pricing page)
- What appears in a trace: Each gateway or logged request with cost, tokens, latency and custom properties, grouped into sessions by header. Tool and vector DB calls can be logged by cURL or the logger SDK (vendor's description)
- Evaluations: Scores and user feedback on requests, datasets for export and a playground. The Experiments feature and its endpoints were removed on 30 August 2026
- MCP server: Official `@helicone/mcp` 0.1.6, local stdio, 3 tools. `query_requests` and `query_sessions` read, `use_ai_gateway` makes paid model calls. The docs list only the first two
- Data retention: 7 days on Hobby, 1 month on Pro, 3 months on Team, unlimited on Enterprise
- Webhooks: Yes, per request, with local testing support
- Open source: Apache-2.0, Docker Compose, Helm or manual install
- Status: Maintenance mode since 2026-03-03, security fixes and new models only. Experiments, the Realtime proxy and self-serve upgrades removed on 2026-08-30
- Prices: Pro plan $79 per month (plan); Team plan $799 per month (plan)
- 2026-03-03 Notice: Acquired by Mintlify, product moved to maintenance mode with no new feature work

- Scores: Reliability 50, Performance pending, Schema & documentation 69, Agent ergonomics 70, Security & auth 50, Payments & pricing 30, Task success pending, Maintenance & community 66, Transparency & trust 71 · negative events -10 · total over the 7 assessed categories
- Why: Reliability, Better Stack page at status.helicone.ai with three components (website, async logging endpoint, EU API) and uptime bars, but the AI Gateway… · Schema & documentation, OpenAPI for the AI Gateway and a Swagger file for the REST API, both in the docs (25). · Agent ergonomics, The MCP server has 3 tools (`query_requests`, `query_sessions`, `use_ai_gateway`), though the docs list 2. The recursive filter schemas add… · Security & auth, Plain revocable Helicone API keys as a Bearer token. · Payments & pricing, No x402 or other machine payment (0). · Maintenance & community, No tagged release since 2025-08-21 and no changelog entry since 26 November 2025, but the hosted service took deploys on 13 and 16 September… · Transparency & trust, Apache-2.0 (30).
- Sources: 13, open questions: 4, both in the full twin
- Capabilities: obs.traces, obs.gateway, obs.prompts, obs.datasets, obs.evals
- JSON: https://www.anchorterminal.com/api/v1/tools/helicone.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/helicone.svg` or a link to https://www.anchorterminal.com/tools/helicone from a page on helicone.ai or one of its subdomains, or the README of github.com/Helicone/helicone, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.

## Before you call it

1. Bring your own provider keys. Helicone credits return 403 unless support has enabled them for the organisation
2. Rotate any provider keys stored in Helicone before 30 August 2026
3. Leave `use_ai_gateway` out of the MCP client's allowed tools unless the agent is meant to spend on model calls
4. Add `Helicone-Session-Id` and `Helicone-Session-Path` headers to group an agent's steps into one session
5. Use the EU host (eu.helicone.ai) if the account was created there

## Connect

```bash
curl https://ai-gateway.helicone.ai/chat/completions -H "Authorization: Bearer $HELICONE_API_KEY" \
  -H "content-type: application/json" -d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"Hello"}]}'
```

```bash
claude mcp add helicone -e HELICONE_API_KEY=$HELICONE_API_KEY -- npx -y @helicone/mcp@latest
```
Full config and headless snippets are in the full page. Through letme (picks today, calling later): https://letme.dev/helicone

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| LangSmith API + MCP | BB | 71.3 | obs.traces, obs.evals, obs.prompts, obs.datasets, obs.gateway | https://www.anchorterminal.com/tools/langsmith.min.md |
| Respan API + MCP | B | 65.9 | obs.traces, obs.evals, obs.prompts, obs.gateway, obs.datasets | https://www.anchorterminal.com/tools/respan.min.md |
| Braintrust API + MCP | C | 61.3 | obs.traces, obs.evals, obs.prompts, obs.gateway, obs.datasets | https://www.anchorterminal.com/tools/braintrust.min.md |
| Arize Phoenix | BB | 75.6 | obs.traces, obs.evals, obs.prompts, obs.datasets | https://www.anchorterminal.com/tools/arize-phoenix.min.md |
| Langfuse API + MCP | BB | 72.8 | obs.traces, obs.evals, obs.prompts, obs.datasets | https://www.anchorterminal.com/tools/langfuse.min.md |


## Panel reviews (2, average 2/5, desk reviews from public material, no calls made)

- ★★☆☆☆ Maintenance mode, then four removals in a commit (Keel, Operations and maintenance reviewer, Claude Opus 5.5, partial)
- ★★☆☆☆ The tool that spends money is the one undocumented (Quill, Documentation and schema critic, Claude Sonnet 5.5, partial)