CrabGlamp

LLM proxy and voice reference

Every supported LLM and voice provider, supported model list per provider with per-token prices, how the spend cap works, proxy base URLs per provider, virtual key auth header shape, and CLI commands for key management. The numbers here are the system of record for billing and quota decisions.

Last updated:

This is the catalog for the LLM proxy and voice surface. For an end-to-end walkthrough, see the Tutorial.

Supported providers and proxy base URLs

The proxy is a transparent pipe — it strips the /openai, /anthropic, or /elevenlabs prefix and forwards the rest to the upstream provider. The OpenAI and Anthropic SDKs don't auto-append /v1, so put it in the base URL. For ElevenLabs the version segment (/v1 or /v2) goes in the request path, so the base URL ends at /elevenlabs.

ProviderProxy base URLAuth header
OpenAIhttps://llm.crabglamp.dev/openai/v1Authorization: Bearer cg-pk-…
Anthropichttps://llm.crabglamp.dev/anthropic/v1x-api-key: cg-pk-… (or Authorization: Bearer cg-pk-…)
ElevenLabshttps://llm.crabglamp.dev/elevenlabsxi-api-key: cg-pk-…

The proxy accepts any of Authorization: Bearer, x-api-key, or xi-api-key — pick whichever is the SDK default for the upstream provider you are calling.

Supported models and prices

Prices are what you pay, per 1M tokens, reviewed monthly. The proxy passes through any model the provider offers; a model not listed below is billed at the provider's published rate.

Chat models — OpenAI:

ModelInput ($/1M tokens)Output ($/1M tokens)
gpt-5.43.7522.50
gpt-5.4-mini1.1256.75
gpt-5.4-nano0.301.875
gpt-5.4-pro45.00270.00
gpt-5.3-codex2.62521.00
gpt-4.13.0012.00
gpt-4.1-mini0.602.40
gpt-4.1-nano0.150.60
gpt-4o3.7515.00
gpt-4o-mini0.2250.90
o33.0012.00
o4-mini1.656.60
o3-mini1.656.60

Chat models — Anthropic:

ModelInput ($/1M tokens)Output ($/1M tokens)
claude-opus-4-67.5037.50
claude-sonnet-4-64.5022.50
claude-opus-4-57.5037.50
claude-sonnet-4-54.5022.50
claude-haiku-4-51.507.50
claude-opus-4-122.50112.50
claude-opus-422.50112.50
claude-sonnet-44.5022.50

Voice — ElevenLabs (per 1k characters):

ModelPrice ($/1k chars)
eleven_v30.18
eleven_multilingual_v20.18
eleven_flash_v2_50.09
eleven_flash_v20.09

Spend-cap math

The cap is the maximum a key may spend across LLM and voice combined. Once your running spend reaches the cap, the next request is rejected with HTTP 429 (Monthly spend limit reached). Enforcement is best-effort: spend is checked against a total that refreshes every couple of minutes, so a request or two may slip through slightly over the cap.

CLI

On-VM CLI for virtual keys:

CommandPurpose
crabglamp keys createCreate this Agent's virtual key and point OpenClaw at the proxy
crabglamp keys statusShow the key's status and this month's spend
crabglamp keys configureSwitch the provider/model the key routes to (OpenClaw)
crabglamp keys regenerateRotate the token; the old token stops working immediately
crabglamp keys revokeRevoke the key (regenerate to restore)

Limits

LimitValue
Minimum spend cap$20
Keys per Agent1
Usage reporting cadencewithin ~2 minutes
Token prefixcg-pk-

Related

View as Markdown — the same content as plain text for AI assistants and offline reading.

Was this helpful?