Compare gateways

Experiential vs Helicone

Helicone is observability with an OpenAI-format gateway, while Experiential is a gateway first with native Anthropic and Gemini APIs, batches and data controls.

Summary

Helicone started as LLM observability: request logs, sessions, properties and prompt management. It added an AI Gateway at ai-gateway.helicone.ai, which takes OpenAI-format requests and offers credits at 0% markup, BYOK, caching and failover. Plans are priced by requests and storage. Helicone's pricing page states that "Helicone is now part of Experiential Labs" (checked 2026-11-19).

import os

from openai import OpenAI

client = OpenAI(
    base_url="https://ai-gateway.helicone.ai",
    api_key=os.environ["HELICONE_API_KEY"],
)
client.chat.completions.create(
    model="gpt-6-luna",
    messages=[{"role": "user", "content": "Hello"}],
    extra_headers={"Helicone-Property-Team": "search", "Helicone-User-Id": "u_42"},
)

Side by side

ExperientialHelicone
HostingHostedHosted, or open source; on-prem on Enterprise
APIsChat, Responses, Messages, Gemini generateContent, embeddings, batchesOpenAI-format chat (Responses, Messages, embeddings, batches not documented)
Model access980 listed models on credits, or your keys100+ providers and models; credits or your keys
PricingList price on tokens, no markup0% markup on credits; plans Hobby free, Pro $79, Team $799 a month, priced on requests and storage
FailoverOrdered waterfall; slow lanes demotedAutomatic failover
Zero data retentionPro: org switch, per request, verdict per responseNot documented
Spend limitsPer-key caps, budgets, tier rate limitsCustom rate limits
Content guardrailsStandard pack or your own classifier"LLM security" listed, no detail
Response cacheNoneYes
Log retentionRequest metadata kept; content under your capture switch7 days (Hobby) to forever (Enterprise)
SSONot yet availableEnterprise (SAML)

Where Experiential is better

  • Native protocols, not just OpenAI format. Claude Code and the Anthropic SDK talk to /v1/messages, and the google-genai SDK to /v1beta, with no translation layer in your code. See Models and the catalog.
  • Data controls. Zero-retention and no-training routing, enforced per org or per request. See Data controls.
  • Spend caps per key, beside budgets and tier rate limits. See Rate limits.
  • Slow providers demoted automatically. See Routing and failover.

When Helicone may fit better

You want observability on calls that stay with your current providers (Helicone's provider proxies and async logging), a response cache, or prompt management.

Migration notes

WhatHeliconeExperiential
Base URLhttps://ai-gateway.helicone.ai, or oai.helicone.ai / anthropic.helicone.ai proxieshttps://api.experientiallabs.ai/v1 (Anthropic SDK: https://api.experientiallabs.ai)
AuthHELICONE_API_KEY, or your provider key plus Helicone-AuthAuthorization: Bearer YOUR_API_KEY only
Model idsModel name, optionally model/provider pins and comma fallbacksPlain slug from GET /v1/models; drop pins and fallbacks
Helicone-Property-*Custom propertiesX-Explabs-Tags (a JSON object, up to 16 labels)
Helicone-User-IdEnd usersafety_identifier in the body
Helicone-Session-IdSessionsX-Explabs-Session-Id
Helicone-Cache-Enabled, Helicone-Prompt-IdCache, promptsNo equivalent
Provider proxiesYour provider key, Helicone observesRuns on Experiential credits unless you connect that key

Switch with an agent

Paste this into your coding agent. It inventories the project first and asks before anything it cannot map.

POST/v1/chat/completions
curl https://oai.helicone.ai/v1/chat/completions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Helicone-Auth: Bearer $HELICONE_API_KEY" \
  -H "Helicone-Session-Id: s_1" \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-6-luna",
       "messages": [{"role": "user", "content": "Hello"}]}'

Sources, checked 2026-11-19

  1. Helicone AI Gateway: base URL, OpenAI SDK format, 0% markup credits, BYOK, caching, failover, rate limits, 100+ providers.
  2. Helicone pricing: plans, retention, SAML SSO and on-prem on Enterprise, and the notice "Helicone is now part of Experiential Labs".
  3. Experiential: /llms.txt, GET /api/models.