Documentation

One base URL — three native protocols. Everything below runs against https://api.olithm.com.

Base URL

https://api.olithm.com/v1

Compatible with every OpenAI SDK. Anthropic clients should point at the same host (the /v1/messages path is native). Codex uses the Responses path. Gemini CLI speaks /v1beta.

Client examples

claude code
# Claude Code — 换 env 即用(/v1/messages 原生橋接)
export ANTHROPIC_BASE_URL=https://api.olithm.com
export ANTHROPIC_AUTH_TOKEN=sk-ov-...
claude
# Codex CLI — Responses API 原生
export OPENAI_API_KEY=sk-ov-...
export OPENAI_BASE_URL=https://api.olithm.com/v1
codex
# Gemini CLI — v1beta 原生
export GEMINI_API_KEY=sk-ov-...
# settings.json → Security →
#   selectedAuthType: "gemini-api-key"
#   baseUrl: "https://api.olithm.com/v1beta"
gemini
openai-chat
from openai import OpenAI
client = OpenAI(api_key="sk-ov-...", base_url="https://api.olithm.com/v1")
import OpenAI from "openai";
const client = new OpenAI({ apiKey: "sk-ov-...", baseURL: "https://api.olithm.com/v1" });

How billing works

Tokens are billed in USD against the published catalog prices. Your reward-tier multiplier (1.00× down to 0.80×) applies to every line item at billing time. Every request appears in the usage log with its exact prices and token counts.

Line itemUnit
Input tokens (uncached)per 1M tokens
Output tokensper 1M tokens
Cache readper 1M tokens
Cache write — 5 min TTLper 1M tokens
Cache write — 1 hour TTLper 1M tokens
Web searchper call
Per-call models (image / video / tasks)per call

Cache-write and web-search prices are per model and only apply when the upstream reports them; models without configured prices fall back to the input price for cache writes.

Reward tiers

Your discount comes from cumulative top-ups, not subscriptions:

Base price1.00×$0+
OV-10.95×$50+
OV-20.90×$200+
OV-30.85×$500+
OV-40.80×$2,000+