// Agent Gateway

Run coding agents on DeepSeek, Qwen, GLM, Kimi, and MiniMax.

XinoAPI gives Codex, OpenClaw, Cline, Roo, and other developer agents a stable gateway for China LLMs: OpenAI-compatible access today, a planned Anthropic-compatible bridge for Claude-style agents, Stripe billing, no China KYC.

Create API key View quickstarts

Supported Agent Clients

Use XinoAPI as the model provider behind developer agents. Keep one billing account while routing tasks across Chinese frontier models.

OpenAI-compatible

Codex CLI

Point Codex at https://api.xinoapi.com/v1 and use deepseek-v4-pro for code planning or deepseek-v4-flash for fast edits.

Bridge planned

Claude Code

Claude Code uses Anthropic's Messages API shape. XinoAPI can support this through an Anthropic-compatible bridge layer for teams that want DeepSeek or Qwen as a coding-agent backend.

Custom provider

OpenClaw

Configure OpenClaw as an OpenAI-compatible provider with a XinoAPI key, then select agent-ready aliases or explicit model IDs.

Quickstarts

Use these snippets as starting points. Replace xino-your-key with an API key from the XinoAPI dashboard.

Codex CLI
export OPENAI_API_KEY="xino-your-key"
export OPENAI_BASE_URL="https://api.xinoapi.com/v1"
export OPENAI_MODEL="deepseek-v4-pro"

codex
Claude Code bridge target
export ANTHROPIC_API_KEY="xino-your-key"
export ANTHROPIC_BASE_URL="https://api.xinoapi.com/anthropic"
export ANTHROPIC_MODEL="deepseek-v4-pro"

claude
OpenClaw
# Provider: OpenAI compatible
# Base URL: https://api.xinoapi.com/v1
# API key: xino-your-key
# Model: deepseek-v4-pro

openclaw configure
Generic OpenAI SDK
from openai import OpenAI

client = OpenAI(
    api_key="xino-your-key",
    base_url="https://api.xinoapi.com/v1",
)

response = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[{"role": "user", "content": "Review this repository."}],
)
Commercial positioning: XinoAPI is not trying to be the cheapest router. It is built for stable coding-agent sessions, China-model coverage, Stripe billing, no China KYC, spend control, and support for production teams.

Recommended Agent Models

Use model choice as a workflow decision: fast edits, complex planning, Chinese codebases, or enterprise reliability.

Model Best For Price Protocol
deepseek-v4-flash Fast code edits, lightweight repo Q&A, low-cost agent loops $0.175/M input, $0.350/M output OpenAI, Anthropic bridge
deepseek-v4-pro Planning, refactors, debugging, multi-file reasoning $0.533/M input, $1.066/M output OpenAI, Anthropic bridge
qwen-plus Chinese codebases, long-context code reading, documentation $0.50/M input, $1.50/M output OpenAI
glm-5.1 Fallback reasoning channel and agent verification passes $1.25/M input, $5.00/M output OpenAI

Agent-Ready Gateway Features

Streaming

Stable long-running tasks

Agent sessions depend on continuous SSE streaming, predictable retries, and clean timeout behavior.

Tools

Function calling support

Use tool calls for search, shell planning, patch generation, structured extraction, and verifier loops.

Billing

Spend control

Minimum $20 top-up, no token discounts, clear usage logs, and enterprise billing for production teams.

Implementation Roadmap

Available Now

  • OpenAI-compatible /v1/chat/completions
  • Streaming responses
  • DeepSeek, Qwen, GLM, Kimi, MiniMax routing
  • Stripe billing and usage accounting

Next Bridge Layer

  • /anthropic/v1/messages bridge for Claude Code
  • Agent aliases like xino-agent-fast and xino-agent-pro
  • Per-agent spend caps and fallback chains
  • Agent logs for enterprise debugging

Why XinoAPI Instead of a Generic Router?

The pitch is reliability and China-model specialization, not lowest token price.

Need XinoAPI Advantage Why It Matters for Agents
China model access DeepSeek, Qwen, GLM, Kimi, MiniMax focus One provider for teams building on Chinese frontier models
No China KYC Stripe and USD billing Overseas developers can start without Chinese phone or payment rails
Agent reliability Routing, fallback, and spend controls Long coding tasks fail less often and are easier to audit
Commercial support Invoice, SLA, dedicated channels Production teams need support paths, not coupon-based pricing