For Codex, Claude Code, and /v1 clients
OneKeyLLM
One API key system for agent workflows: connect Codex, run Claude Code, send /v1 requests, and see remaining spend before a key reaches production.
export ONEKEYLLM_API_KEY="sk-lazytech-..." codex --config model_provider='"onekeyllm"'
Responses API, SSE streaming, and key-level spend checks.
export ANTHROPIC_BASE_URL="https://api.tokapi.ai" export ANTHROPIC_AUTH_TOKEN="sk-lazytech-..." claude
GET /v1/usage
Live key budget
The browser calls /v1/usage directly. OneKeyLLM never stores the key on this site.
One base URL for the agent tools your team already uses.
Integration surface
Two agent protocols, one key and usage model.
Point Codex at /v1/responses, point Claude Code at /v1/messages, then monitor each sk-lazytech key through the same usage API.
Responses API
Codex
Use a dedicated model provider with wire_api = "responses" and ONEKEYLLM_API_KEY as the local bearer source.
Anthropic Messages
Claude Code
Set ANTHROPIC_BASE_URL to the OneKeyLLM API base URL and pass the sk-lazytech key through ANTHROPIC_AUTH_TOKEN.
Budget check
Usage
Read limit_usd, used_usd, and remaining_usd from /v1/usage for the authenticated key.
Agent-compatible endpoints
Responses, Chat Completions, Anthropic Messages, token counting, files, models, and usage live under /v1.
Spend-aware keys
Each sk-lazytech key returns its configured limit, consumed amount, and remaining USD balance.
Static and deployable
The documentation exports as a static Next.js site, while /usage calls the OneKeyLLM API directly through CORS.
Use one key, then wire the agent.
Configure Codex or Claude Code in minutes, then verify remaining budget before handing the key to an automated workflow.
curl https://api.tokapi.ai/v1/responses \
-H "Authorization: Bearer $ONEKEYLLM_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"gpt-5.4","input":"Ship it."}'