LM0
API Reference API Auth

Authentication

How to authenticate to the LM0 public API with OpenAI or Anthropic header styles.

On this page (9)

LM0 accepts both common client authentication styles on every /v1/* route so you can keep stock SDKs unchanged.

Header styles

OpenAI style

HTTP 2 lines 
Authorization: Bearer lm0_...

Anthropic style

HTTP 3 lines 
x-api-key: lm0_...
anthropic-version: 2023-06-01

Both styles resolve the same API key. Prefer the style your SDK already sends.

Examples

Bash 9 lines 
# OpenAI-style
curl https://api.17.wtf/v1/models \
  -H "Authorization: Bearer $LM0_API_KEY"

# Anthropic-style
curl https://api.17.wtf/v1/models \
  -H "x-api-key: $LM0_API_KEY" \
  -H "anthropic-version: 2023-06-01"

Runtime modes

DB-backed production mode

When the gateway is configured with a database:

  • Keys resolve to user records
  • Banned accounts are blocked
  • Per-user RPM limits apply by account tier
  • Package token balance is enforced for generation
  • Pool / model restrictions and beta access are enforced

Legacy / local mode

Without a database block:

  • Keys come from virtual_keys in config (optional model allow-list and RPM)
  • If no virtual keys are configured, the gateway may run open (local/dev only)
Warning

Never run open (no-auth) mode on a public network. Use DB-backed keys in production.

Keys

Keys are typically prefixed (for example lm0_…). Create and revoke keys from the product console when available, or via admin APIs in DB mode.

Errors

Common auth-related outcomes:

Situation Typical result
Missing / invalid key 401
Banned account 403
Insufficient credits (DB mode) 402
Rate limited (RPM) 429

Exact bodies follow the gateway’s error shape for the surface you called.

LM0 · Documentation for every model you ship

api.17.wtf