# Inferris API: plug your agent in Inferris serves open models (today: `glm-5.3`) behind OpenAI- and Anthropic-compatible endpoints. Point your agent at the base URL, give it a key, pick the model. Nothing else changes. Human-readable version: https://inferris.ai/developers ## The settings | Setting | Value | | ----------------------------- | ------------------------------------------------------------------------------------- | | OpenAI-compatible base URL | `https://api.inferris.ai/v1` | | Anthropic-compatible base URL | `https://api.inferris.ai` (clients add `/v1/messages`) | | Authentication | `Authorization: Bearer $INFERRIS_API_KEY` (on `/v1/messages`, `x-api-key` also works) | | Model ID | `glm-5.3` (`GET /v1/models` with your key lists every model it can call) | | Context window | Not published yet | | Max output tokens | Not published yet | `https://inferris.ai/v1` is the same gateway as `https://api.inferris.ai/v1`. Get a key: sign in at https://inferris.ai and open API keys (https://inferris.ai/keys). Keep it in an environment variable; anyone with the key can spend your balance. ``` export INFERRIS_API_KEY="sk-..." ``` ## Endpoints | Endpoint | API shape | Used by | | --------------------------- | ------------------------------------------------------ | -------------------------------------------- | | `POST /v1/chat/completions` | OpenAI Chat Completions (stream with `"stream": true`) | OpenAI SDKs, OpenCode, Cline, Continue, curl | | `POST /v1/responses` | OpenAI Responses | Codex CLI | | `POST /v1/messages` | Anthropic Messages | Claude Code, Anthropic SDKs | | `GET /v1/models` | Models your key can call | Any client | ## Connect your agent ### Claude Code Uses the Anthropic Messages endpoint. Add these to your shell profile, then run `claude`. ``` export ANTHROPIC_BASE_URL="https://api.inferris.ai" export ANTHROPIC_AUTH_TOKEN="$INFERRIS_API_KEY" export ANTHROPIC_MODEL="glm-5.3" export ANTHROPIC_DEFAULT_OPUS_MODEL="glm-5.3" export ANTHROPIC_DEFAULT_SONNET_MODEL="glm-5.3" export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-5.3" ``` ### Codex CLI Uses the OpenAI Responses endpoint. Add to `~/.codex/config.toml` and export `INFERRIS_API_KEY`. ``` model = "glm-5.3" model_provider = "inferris" [model_providers.inferris] name = "Inferris" base_url = "https://api.inferris.ai/v1" wire_api = "responses" env_key = "INFERRIS_API_KEY" ``` ### OpenCode Uses the OpenAI Chat Completions endpoint. Add a provider to `opencode.json` (project) or `~/.config/opencode/opencode.json`. Format from the OpenCode documentation; not yet run end to end against Inferris. ``` { "$schema": "https://opencode.ai/config.json", "provider": { "inferris": { "npm": "@ai-sdk/openai-compatible", "name": "Inferris", "options": { "baseURL": "https://api.inferris.ai/v1", "apiKey": "{env:INFERRIS_API_KEY}" }, "models": { "glm-5.3": { "name": "glm-5.3" } } } } } ``` ### Cline, Roo Code, Continue and other OpenAI-compatible tools Choose the "OpenAI Compatible" provider (the name varies by tool) and fill in three fields. Field names differ between tools; we have not tested each tool individually. ``` Base URL: https://api.inferris.ai/v1 API key: Model ID: glm-5.3 ``` ## Call it from code ### curl (OpenAI) ``` curl https://api.inferris.ai/v1/chat/completions \ -H "Authorization: Bearer $INFERRIS_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "glm-5.3", "messages": [{"role": "user", "content": "Say hello in one sentence."}] }' ``` ### Python (OpenAI) ``` # pip install openai import os from openai import OpenAI client = OpenAI( base_url="https://api.inferris.ai/v1", api_key=os.environ["INFERRIS_API_KEY"], ) reply = client.chat.completions.create( model="glm-5.3", messages=[{"role": "user", "content": "Say hello in one sentence."}], ) print(reply.choices[0].message.content) ``` ### Node.js (OpenAI) ``` // npm install openai import OpenAI from 'openai' const client = new OpenAI({ baseURL: 'https://api.inferris.ai/v1', apiKey: process.env.INFERRIS_API_KEY, }) const reply = await client.chat.completions.create({ model: 'glm-5.3', messages: [{ role: 'user', content: 'Say hello in one sentence.' }], }) console.log(reply.choices[0].message.content) ``` ### curl (Anthropic) ``` curl https://api.inferris.ai/v1/messages \ -H "Authorization: Bearer $INFERRIS_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "glm-5.3", "max_tokens": 512, "messages": [{"role": "user", "content": "Say hello in one sentence."}] }' ``` ### Python (Anthropic) ``` # pip install anthropic import os from anthropic import Anthropic client = Anthropic( base_url="https://api.inferris.ai", api_key=os.environ["INFERRIS_API_KEY"], ) reply = client.messages.create( model="glm-5.3", max_tokens=512, messages=[{"role": "user", "content": "Say hello in one sentence."}], ) print(reply.content[0].text) ``` ### Node.js (Anthropic) ``` // npm install @anthropic-ai/sdk import Anthropic from '@anthropic-ai/sdk' const client = new Anthropic({ baseURL: 'https://api.inferris.ai', apiKey: process.env.INFERRIS_API_KEY, }) const reply = await client.messages.create({ model: 'glm-5.3', max_tokens: 512, messages: [{ role: 'user', content: 'Say hello in one sentence.' }], }) console.log(reply.content[0].type === 'text' ? reply.content[0].text : '') ``` ## Errors and retries Errors come back as JSON with `error.message` and `error.type`. Branch on the HTTP status code; some messages are still in Chinese. | Status | Meaning | Retry? | What to do | | --------------- | -------------------------------------------------------------------------------------------- | ------ | --------------------------------------------------------------------------------------------------- | | 400 | The request body is invalid. | No | Fix the request. The same body will fail again. | | 401 | Key missing, wrong, disabled or expired. | No | Check the key. | | 403 | Balance or key quota used up (`insufficient_user_quota`), or the key may not use this model. | No | Top up in Wallet or raise the key limit. | | 429 | Rate limit reached, or the model is at capacity. | Yes | Wait for `Retry-After` (seconds) when present; otherwise back off 1, 2, 4 … up to 30 s with jitter. | | 503 | No capacity is available for this model right now. | Yes | Retry with backoff; check https://inferris.ai/status if it lasts. | | 500 / 502 / 504 | The gateway or the model backend failed or timed out. | Yes | Retry with backoff. | Live rate limits: https://inferris.ai/docs/rate-limits ## More - Agent setup instructions: https://inferris.ai/agents.md - Site index for language models: https://inferris.ai/llms.txt - Models and prices: https://inferris.ai/pricing - Status: https://inferris.ai/status