> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gravitygtm.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Helium API error responses and per-key rate limits

> Helium shares Horizon's error shape and status codes; per-key rate limits and per-tenant shared credits, and why DELETE /v1/cache doesn't apply to Helium's endpoints.

Helium shares Horizon's exact error envelope, HTTP status codes, and
`request_id` convention — see
[Horizon's Errors & Rate Limits page](/horizon/errors-and-rate-limits) for
the full shape and status-code table (`400`/`401`/`402`/`422`/`429`/`502`/`503`).
The same envelope applies to every Helium REST and MCP call.

## What's different for Helium

**No cache-eviction support.** Horizon's `DELETE /v1/cache` endpoint does
not apply to any Helium endpoint — Helium's research results aren't cached
in the first place (see [Credits & Pricing](/helium/credits-and-pricing)),
so there's nothing to evict.

**All billed endpoints charge on a miss.** Helium has no equivalent to
Horizon's "charged on success only" `person.email` exception — every one
of `search_investors`, `enrich_investor`, and `research_company` charges
its full listed price even on a clean "nothing found" result.

## Rate limits

Rate limits apply **per API key**, using the same tiers as Horizon (10
requests/minute on a free-trial key, 60/minute on any paid pack). If your
workspace has installed both Horizon's and Helium's Slack apps, each app's
own dispatch key has its **own independent rate-limit bucket** — hammering
one doesn't eat into the other's. **Credits, however, are shared
tenant-wide** across every key and both products — a rate limit is per-key,
a credit balance is per-account.