Documentation

API Reference & Setup

One API key for Claude Code, OpenCode, OpenClaw, Codex CLI, Gemini CLI, Cursor, and direct API access. Claude & GPT models at 50% off official pricing -- pay as you go, no subscription required.

Table of Contents

1. Getting Started

Zenn.Engineering is a drop-in API proxy for Anthropic, OpenAI, and Google AI models. You use a single ck_-prefixed API key across all supported tools and SDKs. No code changes required — just set the base URL and API key.

1. Get an API Key

Top up credits, then create a key from your dashboard.

2. Set Base URL

Point your tool to https://zenn.engineering/api/v1

3. Start Coding

Works instantly with Claude Code, OpenCode, OpenClaw, and more.

Base URLs

Provider	Base URL
Claude (Anthropic)	https://zenn.engineering/api/v1
Codex (OpenAI)	https://zenn.engineering/api/v1/codex
Gemini (Google)	https://zenn.engineering/api/v1/gemini

2. Claude Code

Anthropic's official CLI for Claude. Set two environment variables and it works as a drop-in replacement — same CLI, same models, 50% off official pricing.

Step 1: Set environment variables

Add to your shell profile (~/.zshrc or ~/.bashrc):

shell

export ANTHROPIC_BASE_URL=https://zenn.engineering/api/v1
export ANTHROPIC_API_KEY=ck_YOUR_API_KEY

Step 2: Restart terminal & run

terminal

# Default model (Sonnet 4.5)
claude

# Use Opus 4.6
claude --model claude-opus-4-6

How it works

Claude Code sends the API key via the x-api-key header (Anthropic SDK native behavior) and appends /messages to the base URL automatically. The anthropic-version and anthropic-beta headers are forwarded to the upstream API.

3. OpenCode

Multi-provider AI coding agent. Configure once — access Claude, Codex, and Gemini models through one JSON config.

Step 1: Install

terminal

npm i -g opencode-ai

Step 2: Create config

Create or edit ~/.config/opencode/opencode.json:

opencode.json

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "anthropic": {
      "options": {
        "baseURL": "https://zenn.engineering/api/v1",
        "apiKey": "ck_YOUR_API_KEY"
      },
      "models": {
        "claude-sonnet-4-6": { "name": "Claude Sonnet 4.6" },
        "claude-opus-4-6": { "name": "Claude Opus 4.6" }
      }
    },
    "zenn-codex": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Zenn Codex",
      "options": {
        "baseURL": "https://zenn.engineering/api/v1/codex",
        "apiKey": "ck_YOUR_API_KEY"
      },
      "models": {
        "gpt-5.3-codex": { "name": "GPT-5.3 Codex" }
      }
    },
    "zenn-gemini": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Zenn Gemini",
      "options": {
        "baseURL": "https://zenn.engineering/api/v1/gemini",
        "apiKey": "ck_YOUR_API_KEY"
      },
      "models": {
        "gemini-3-pro-preview": { "name": "Gemini 3.0 Pro" },
        "gemini-3-flash-preview": { "name": "Gemini 3.0 Flash" }
      }
    }
  }
}

Step 3: Start coding

terminal

cd your-project
opencode

# Switch models with /model

How it works

OpenCode sends the API key via the Authorization: Bearer header. The anthropic provider uses the native Anthropic SDK format, while zenn-codex and zenn-gemini use the OpenAI-compatible provider adapter.

4. OpenClaw

Open-source autonomous AI coding agent. Configure Zenn as a custom provider for Claude, Codex, and Gemini access through one config file.

Step 1: Install

terminal

curl -fsSL https://openclaw.ai/install.sh | bash

Step 2: Configure providers

Create or edit ~/.openclaw/openclaw.json:

openclaw.json

{
  "models": {
    "providers": {
      "zenn-claude": {
        "baseUrl": "https://zenn.engineering/api/v1",
        "apiKey": "ck_YOUR_API_KEY",
        "api": "anthropic-messages",
        "models": [
          { "id": "claude-sonnet-4-6", "name": "Claude Sonnet 4.6" },
          { "id": "claude-opus-4-6", "name": "Claude Opus 4.6" }
        ]
      },
      "zenn-codex": {
        "baseUrl": "https://zenn.engineering/api/v1/codex",
        "apiKey": "ck_YOUR_API_KEY",
        "api": "openai-responses",
        "models": [
          { "id": "gpt-5.3-codex", "name": "GPT-5.3 Codex" }
        ]
      },
      "zenn-gemini": {
        "baseUrl": "https://zenn.engineering/api/v1/gemini",
        "apiKey": "ck_YOUR_API_KEY",
        "api": "openai-completions",
        "models": [
          { "id": "gemini-3-pro-preview", "name": "Gemini 3.0 Pro" },
          { "id": "gemini-3-flash-preview", "name": "Gemini 3.0 Flash" }
        ]
      }
    }
  }
}

Step 3: Start coding

terminal

cd your-project
openclaw

API format reference

Provider	`api` value	Description
Claude	anthropic-messages	Anthropic Messages API format
Codex	openai-responses	OpenAI Responses API format
Gemini	openai-completions	OpenAI Chat Completions format

5. Codex CLI

OpenAI's Codex CLI for GPT-5.x code generation, routed through Zenn.Engineering's /v1/responses endpoint at 50% off official pricing. Codex CLI speaks the OpenAI Responses API by default, so configure it as a custom provider.

Step 1 — Install Codex CLI

shell

npm install -g @openai/codex
# or:  brew install codex

Step 2 — Generate an API key

At zenn.engineering/manage-api-keys, create a key. It must start with ck_. OpenAI's sk-… keys will not work.

Step 3 — Write ~/.codex/config.toml

~/.codex/config.toml

model = "gpt-5-codex"
model_provider = "zenn"

[model_providers.zenn]
name = "Zenn"
base_url = "https://zenn.engineering/api/v1"
env_key = "ZENN_API_KEY"
wire_api = "responses"

Important: base_url stops at /api/v1 (no trailing slash, no /responses). Codex CLI appends the path itself. wire_api = "responses" is required.

Step 4 — Export the API key

shell

export ZENN_API_KEY="ck_your_key_here"

# To persist across shells, append to your rc file:
echo 'export ZENN_API_KEY="ck_your_key_here"' >> ~/.zshrc   # or ~/.bashrc

The env var name (ZENN_API_KEY) must match env_key in the config.

Step 5 — Run

terminal

codex
# or for one-shot:
codex exec "refactor this function"

Verify the endpoint is reachable

shell

curl https://zenn.engineering/api/v1/responses

Returns JSON with the supported OpenAI-format models. If this fails, Codex CLI cannot connect either.

Troubleshooting

401 unauthorized: key doesn't start with ck_, or $ZENN_API_KEY is empty (run echo $ZENN_API_KEY to verify).
missing env ZENN_API_KEY: export it in the shell where you launch codex, or append to ~/.zshrc and source it.
stuck on "connecting": wire_api missing or set to anything other than "responses".
MCP client failed to start: pre-existing MCP server config in ~/.codex/config.toml. Usually a warning and safe to ignore; to silence it, add mcp_servers = {} to the config.
402 insufficient credits: top up at zenn.engineering/settings. Minimum balance to call the API is 200 credits ($2.00).

6. Gemini CLI

Google's Gemini CLI with full streaming support, routed through Zenn.Engineering at discounted pricing.

Set environment variables

shell

export GEMINI_API_BASE_URL=https://zenn.engineering/api/v1/gemini
export GEMINI_API_KEY=ck_YOUR_API_KEY

Run

terminal

gemini --model gemini-3-pro-preview "explain this code"

7. Cursor IDE

Use Claude models in Cursor IDE via Zenn.Engineering.

Configure in Cursor Settings

Go to Cursor Settings → Models → Anthropic

Anthropic API Key

ck_YOUR_API_KEY

Override Anthropic Base URL

https://zenn.engineering/api/v1

Available models

Select claude-sonnet-4-6, claude-haiku-4-5, or claude-opus-4-6 in the model picker.

8. Direct API Usage

The Zenn.Engineering API is fully compatible with the Anthropic Messages API. Use it directly with cURL, the Anthropic SDK (Node.js/Python), or any HTTP client.

cURL

bash

curl -X POST https://zenn.engineering/api/v1/messages \
  -H "Authorization: Bearer ck_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": "Hello, Claude!"}
    ]
  }'

Node.js / TypeScript

typescript

import Anthropic from '@anthropic-ai/sdk';

const client = new Anthropic({
  apiKey: 'ck_YOUR_API_KEY',
  baseURL: 'https://zenn.engineering/api/v1',
});

const message = await client.messages.create({
  model: 'claude-sonnet-4-6',
  max_tokens: 1024,
  messages: [
    { role: 'user', content: 'Hello, Claude!' }
  ],
});

Python

python

import anthropic

client = anthropic.Anthropic(
    api_key="ck_YOUR_API_KEY",
    base_url="https://zenn.engineering/api/v1",
)

message = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Hello, Claude!"}
    ],
)

9. Models & Pricing

All pricing is credit-based (100 credits = $1.00 USD). Prices shown are per million tokens (MTok). Claude & GPT models are priced at 50% of official pricing (50% off). Prices match the effective rate of Claude Max ($200/mo) and Codex Pro plans — pay as you go, no subscription required.

Claude (Anthropic) — 50% off official

Model ID	Input / MTok	Output / MTok	Savings	Tier
claude-sonnet-4-6	$1.50	$7.50	50% off	Starter
claude-haiku-4-5	$0.40	$2.00	50% off	Starter
claude-opus-4-6	$7.50	$37.50	50% off	Opus

OpenAI / GPT — 50% off official

Model ID	Input / MTok	Output / MTok	Savings	Tier
gpt-5	$1.25	$5.00	50% off	Starter
gpt-5.1	$1.25	$5.00	50% off	Starter
gpt-5.2	$2.50	$7.50	50% off	Starter
gpt-5-codex	$1.25	$5.00	50% off	Starter
gpt-5-codex-mini	$0.15	$0.63	50% off	Starter
gpt-5.1-codex	$1.25	$5.00	50% off	Starter
gpt-5.1-codex-mini	$0.15	$0.63	50% off	Starter
gpt-5.1-codex-max	$1.25	$5.00	50% off	Starter
gpt-5.2-codex	$2.50	$7.50	50% off	Starter
gpt-5.3-codex	$2.50	$7.50	50% off	Starter
gpt-5.3-codex-spark	$2.50	$7.50	50% off	Starter

Gemini (Google) — 10% off official

Model ID	Input / MTok	Output / MTok	Savings	Tier
gemini-3-pro-official	$1.76	$10.56	10% off	Starter
gemini-3-pro-preview-official	$1.76	$10.56	10% off	Starter
gemini-3-flash-official	$0.44	$2.64	10% off	Starter
gemini-3-flash-preview-official	$0.44	$2.64	10% off	Starter
gemini-3.1-pro	$0.05	$0.30	Nominal	Starter
gemini-3.1-pro-preview-official	$1.76	$10.56	10% off	Starter
gemini-3.1-fast	$0.55	$3.30	Low cost	Starter
gemini-3.1-thinking	$0.55	$3.30	Low cost	Starter
gemini-3.1-flash-lite-preview-official	$0.22	$1.32	Low cost	Starter
gemini-2.5-pro-official	$1.10	$8.80	10% off	Starter
gemini-2.5-flash-official	$0.26	$2.20	Low cost	Starter
gemini-2.5-flash-lite-official	$0.09	$0.35	Ultra low	Starter
gemini-2.0-flash-official	$0.13	$0.53	Ultra low	Starter
gemini-2.0-flash-lite-official	$0.07	$0.26	Ultra low	Starter

Other Models (DeepSeek, Qwen, GLM, Kimi, MiniMax)

Model ID	Provider	Input / MTok	Output / MTok	Tier
deepseek-v3.2	DeepSeek	$0.44	$1.76	Starter
glm-5	Zhipu	$0.44	$1.98	Starter
qwen3.5-plus	Alibaba	$0.44	$2.64	Starter
qwen3.5-flash	Alibaba	$0.13	$1.32	Starter
qwen3-max	Alibaba	$0.77	$3.08	Starter
kimi-k2.5	Moonshot	$0.44	$2.31	Starter
MiniMax-M2.5	MiniMax	$0.23	$0.92	Starter

Free / Promotional

Model ID	Input / MTok	Output / MTok	Tier
nano-banana-2	$0.05	$0.30	Starter

10. Authentication

All API keys use the ck_ prefix. The proxy accepts multiple authentication header formats:

Header	Format	Used By
x-api-key	ck_...	Claude Code, Anthropic SDK
Authorization	Bearer ck_...	OpenCode, OpenAI SDK, cURL
anthropic-api-key	ck_...	Alternative Anthropic header
x-goog-api-key	ck_...	Gemini CLI compatibility

Forwarded headers

The proxy forwards anthropic-version (defaults to 2023-06-01) and anthropic-beta headers to the upstream Anthropic API. Streaming is fully supported via Server-Sent Events (SSE).

11. Rate Limits & Errors

Rate limits

Requests per hour

1,000

Minimum credit balance

200 credits ($2.00)

Streaming timeout

300s (5 min)

Rate limit info is returned in response headers: x-ratelimit-limit, x-ratelimit-remaining, x-ratelimit-reset.

Error responses

All errors return JSON:

json

{
  "error": {
    "type": "error_type",
    "message": "Human-readable description"
  }
}

Status	Type	Cause
400	invalid_request_error	Missing required parameters
401	authentication_error	Invalid or missing API key
402	insufficient_credits	Credit balance below $2.00
403	access_denied	Model not available at your tier
429	rate_limit_error	Rate limit exceeded (1,000/hr)

12. Tiers

Tiers are determined by cumulative top-up amount. Higher tiers unlock additional models.

Starter

$19.99+ cumulative top-up

Unlocks:

Claude Sonnet 4.6
GPT-5.2 / GPT-5.3 Codex
Gemini 3.0 Pro & Flash
API key creation

Opus

$99.99+ cumulative top-up

Everything in Starter, plus:

Claude Opus 4.6
Priority queue

13. Audio Generation (Fish Audio)

Text-to-speech, voice cloning, and speech recognition powered by Fish Audio. Requires Starter tier ($19.99+).

Endpoint

POST https://zenn.engineering/api/v1/audio/generations
GET  https://zenn.engineering/api/v1/audio/generations   (list models)

Available Models

Model ID	Type	Credits	Price	Required Input
audio-tts	Text to Speech	2	$0.020	text
audio-clone	Voice Clone	2	$0.020	text + audio
audio-asr	Speech Recognition	1	$0.010	audio

Text to Speech

cURL

curl -X POST https://zenn.engineering/api/v1/audio/generations \
  -H "Authorization: Bearer ck_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "audio-tts",
    "text": "Hello, welcome to Zenn!",
    "format": "mp3"
  }'

Optional parameters: format (mp3, wav, opus), temperature, reference_id (existing voice model ID).

Voice Clone

cURL

curl -X POST https://zenn.engineering/api/v1/audio/generations \
  -H "Authorization: Bearer ck_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "audio-clone",
    "text": "Say this in the cloned voice",
    "audio": "https://example.com/reference-voice.mp3"
  }'

The audio field is a URL to the reference voice sample for zero-shot cloning.

Speech Recognition (ASR)

cURL

curl -X POST https://zenn.engineering/api/v1/audio/generations \
  -H "Authorization: Bearer ck_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "audio-asr",
    "audio": "https://example.com/speech.mp3",
    "language": "en"
  }'

Response Format

TTS / Voice Clone returns an audio URL:

TTS Response

{
  "model": "audio-tts",
  "created": 1709654321,
  "data": [{ "url": "https://storage.zenn.run/generated/audio/..." }]
}

ASR returns transcribed text:

ASR Response

{
  "model": "audio-asr",
  "created": 1709654321,
  "text": "Hello, welcome to Zenn!",
  "duration": 2.5
}

Ready to start?

One key works across Claude Code, OpenCode, OpenClaw, Codex CLI, Gemini CLI, and Cursor. Top up credits and create your API key.

Get Credits Manage API Keys

Documentation

API Reference & Setup

One API key for Claude Code, OpenCode, OpenClaw, Codex CLI, Gemini CLI, Cursor, and direct API access. Claude & GPT models at 50% off official pricing -- pay as you go, no subscription required.

Table of Contents

1. Getting Started

1. Get an API Key

Top up credits, then create a key from your dashboard.

2. Set Base URL

Point your tool to https://zenn.engineering/api/v1

3. Start Coding

Works instantly with Claude Code, OpenCode, OpenClaw, and more.

Base URLs

Provider	Base URL
Claude (Anthropic)	https://zenn.engineering/api/v1
Codex (OpenAI)	https://zenn.engineering/api/v1/codex
Gemini (Google)	https://zenn.engineering/api/v1/gemini

2. Claude Code

Anthropic's official CLI for Claude. Set two environment variables and it works as a drop-in replacement — same CLI, same models, 50% off official pricing.

Step 1: Set environment variables

Add to your shell profile (~/.zshrc or ~/.bashrc):

shell

export ANTHROPIC_BASE_URL=https://zenn.engineering/api/v1
export ANTHROPIC_API_KEY=ck_YOUR_API_KEY

Step 2: Restart terminal & run

terminal

# Default model (Sonnet 4.5)
claude

# Use Opus 4.6
claude --model claude-opus-4-6

How it works

3. OpenCode

Multi-provider AI coding agent. Configure once — access Claude, Codex, and Gemini models through one JSON config.

Step 1: Install

terminal

npm i -g opencode-ai

Step 2: Create config

Create or edit ~/.config/opencode/opencode.json:

opencode.json

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "anthropic": {
      "options": {
        "baseURL": "https://zenn.engineering/api/v1",
        "apiKey": "ck_YOUR_API_KEY"
      },
      "models": {
        "claude-sonnet-4-6": { "name": "Claude Sonnet 4.6" },
        "claude-opus-4-6": { "name": "Claude Opus 4.6" }
      }
    },
    "zenn-codex": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Zenn Codex",
      "options": {
        "baseURL": "https://zenn.engineering/api/v1/codex",
        "apiKey": "ck_YOUR_API_KEY"
      },
      "models": {
        "gpt-5.3-codex": { "name": "GPT-5.3 Codex" }
      }
    },
    "zenn-gemini": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Zenn Gemini",
      "options": {
        "baseURL": "https://zenn.engineering/api/v1/gemini",
        "apiKey": "ck_YOUR_API_KEY"
      },
      "models": {
        "gemini-3-pro-preview": { "name": "Gemini 3.0 Pro" },
        "gemini-3-flash-preview": { "name": "Gemini 3.0 Flash" }
      }
    }
  }
}

Step 3: Start coding

terminal

cd your-project
opencode

# Switch models with /model

How it works

4. OpenClaw

Open-source autonomous AI coding agent. Configure Zenn as a custom provider for Claude, Codex, and Gemini access through one config file.

Step 1: Install

terminal

curl -fsSL https://openclaw.ai/install.sh | bash

Step 2: Configure providers

Create or edit ~/.openclaw/openclaw.json:

openclaw.json

{
  "models": {
    "providers": {
      "zenn-claude": {
        "baseUrl": "https://zenn.engineering/api/v1",
        "apiKey": "ck_YOUR_API_KEY",
        "api": "anthropic-messages",
        "models": [
          { "id": "claude-sonnet-4-6", "name": "Claude Sonnet 4.6" },
          { "id": "claude-opus-4-6", "name": "Claude Opus 4.6" }
        ]
      },
      "zenn-codex": {
        "baseUrl": "https://zenn.engineering/api/v1/codex",
        "apiKey": "ck_YOUR_API_KEY",
        "api": "openai-responses",
        "models": [
          { "id": "gpt-5.3-codex", "name": "GPT-5.3 Codex" }
        ]
      },
      "zenn-gemini": {
        "baseUrl": "https://zenn.engineering/api/v1/gemini",
        "apiKey": "ck_YOUR_API_KEY",
        "api": "openai-completions",
        "models": [
          { "id": "gemini-3-pro-preview", "name": "Gemini 3.0 Pro" },
          { "id": "gemini-3-flash-preview", "name": "Gemini 3.0 Flash" }
        ]
      }
    }
  }
}

Step 3: Start coding

terminal

cd your-project
openclaw

API format reference

Provider	`api` value	Description
Claude	anthropic-messages	Anthropic Messages API format
Codex	openai-responses	OpenAI Responses API format
Gemini	openai-completions	OpenAI Chat Completions format

5. Codex CLI

Step 1 — Install Codex CLI

shell

npm install -g @openai/codex
# or:  brew install codex

Step 2 — Generate an API key

At zenn.engineering/manage-api-keys, create a key. It must start with ck_. OpenAI's sk-… keys will not work.

Step 3 — Write ~/.codex/config.toml

~/.codex/config.toml

model = "gpt-5-codex"
model_provider = "zenn"

[model_providers.zenn]
name = "Zenn"
base_url = "https://zenn.engineering/api/v1"
env_key = "ZENN_API_KEY"
wire_api = "responses"

Important: base_url stops at /api/v1 (no trailing slash, no /responses). Codex CLI appends the path itself. wire_api = "responses" is required.

Step 4 — Export the API key

shell

export ZENN_API_KEY="ck_your_key_here"

# To persist across shells, append to your rc file:
echo 'export ZENN_API_KEY="ck_your_key_here"' >> ~/.zshrc   # or ~/.bashrc

The env var name (ZENN_API_KEY) must match env_key in the config.

Step 5 — Run

terminal

codex
# or for one-shot:
codex exec "refactor this function"

Verify the endpoint is reachable

shell

curl https://zenn.engineering/api/v1/responses

Returns JSON with the supported OpenAI-format models. If this fails, Codex CLI cannot connect either.

Troubleshooting

401 unauthorized: key doesn't start with ck_, or $ZENN_API_KEY is empty (run echo $ZENN_API_KEY to verify).
missing env ZENN_API_KEY: export it in the shell where you launch codex, or append to ~/.zshrc and source it.
stuck on "connecting": wire_api missing or set to anything other than "responses".
MCP client failed to start: pre-existing MCP server config in ~/.codex/config.toml. Usually a warning and safe to ignore; to silence it, add mcp_servers = {} to the config.
402 insufficient credits: top up at zenn.engineering/settings. Minimum balance to call the API is 200 credits ($2.00).

6. Gemini CLI

Google's Gemini CLI with full streaming support, routed through Zenn.Engineering at discounted pricing.

Set environment variables

shell

export GEMINI_API_BASE_URL=https://zenn.engineering/api/v1/gemini
export GEMINI_API_KEY=ck_YOUR_API_KEY

Run

terminal

gemini --model gemini-3-pro-preview "explain this code"

7. Cursor IDE

Use Claude models in Cursor IDE via Zenn.Engineering.

Configure in Cursor Settings

Go to Cursor Settings → Models → Anthropic

Anthropic API Key

ck_YOUR_API_KEY

Override Anthropic Base URL

https://zenn.engineering/api/v1

Available models

Select claude-sonnet-4-6, claude-haiku-4-5, or claude-opus-4-6 in the model picker.

8. Direct API Usage

The Zenn.Engineering API is fully compatible with the Anthropic Messages API. Use it directly with cURL, the Anthropic SDK (Node.js/Python), or any HTTP client.

cURL

bash

curl -X POST https://zenn.engineering/api/v1/messages \
  -H "Authorization: Bearer ck_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": "Hello, Claude!"}
    ]
  }'

Node.js / TypeScript

typescript

import Anthropic from '@anthropic-ai/sdk';

const client = new Anthropic({
  apiKey: 'ck_YOUR_API_KEY',
  baseURL: 'https://zenn.engineering/api/v1',
});

const message = await client.messages.create({
  model: 'claude-sonnet-4-6',
  max_tokens: 1024,
  messages: [
    { role: 'user', content: 'Hello, Claude!' }
  ],
});

Python

python

import anthropic

client = anthropic.Anthropic(
    api_key="ck_YOUR_API_KEY",
    base_url="https://zenn.engineering/api/v1",
)

message = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Hello, Claude!"}
    ],
)

9. Models & Pricing

Claude (Anthropic) — 50% off official

Model ID	Input / MTok	Output / MTok	Savings	Tier
claude-sonnet-4-6	$1.50	$7.50	50% off	Starter
claude-haiku-4-5	$0.40	$2.00	50% off	Starter
claude-opus-4-6	$7.50	$37.50	50% off	Opus

OpenAI / GPT — 50% off official

Model ID	Input / MTok	Output / MTok	Savings	Tier
gpt-5	$1.25	$5.00	50% off	Starter
gpt-5.1	$1.25	$5.00	50% off	Starter
gpt-5.2	$2.50	$7.50	50% off	Starter
gpt-5-codex	$1.25	$5.00	50% off	Starter
gpt-5-codex-mini	$0.15	$0.63	50% off	Starter
gpt-5.1-codex	$1.25	$5.00	50% off	Starter
gpt-5.1-codex-mini	$0.15	$0.63	50% off	Starter
gpt-5.1-codex-max	$1.25	$5.00	50% off	Starter
gpt-5.2-codex	$2.50	$7.50	50% off	Starter
gpt-5.3-codex	$2.50	$7.50	50% off	Starter
gpt-5.3-codex-spark	$2.50	$7.50	50% off	Starter

Gemini (Google) — 10% off official

Model ID	Input / MTok	Output / MTok	Savings	Tier
gemini-3-pro-official	$1.76	$10.56	10% off	Starter
gemini-3-pro-preview-official	$1.76	$10.56	10% off	Starter
gemini-3-flash-official	$0.44	$2.64	10% off	Starter
gemini-3-flash-preview-official	$0.44	$2.64	10% off	Starter
gemini-3.1-pro	$0.05	$0.30	Nominal	Starter
gemini-3.1-pro-preview-official	$1.76	$10.56	10% off	Starter
gemini-3.1-fast	$0.55	$3.30	Low cost	Starter
gemini-3.1-thinking	$0.55	$3.30	Low cost	Starter
gemini-3.1-flash-lite-preview-official	$0.22	$1.32	Low cost	Starter
gemini-2.5-pro-official	$1.10	$8.80	10% off	Starter
gemini-2.5-flash-official	$0.26	$2.20	Low cost	Starter
gemini-2.5-flash-lite-official	$0.09	$0.35	Ultra low	Starter
gemini-2.0-flash-official	$0.13	$0.53	Ultra low	Starter
gemini-2.0-flash-lite-official	$0.07	$0.26	Ultra low	Starter

Other Models (DeepSeek, Qwen, GLM, Kimi, MiniMax)

Model ID	Provider	Input / MTok	Output / MTok	Tier
deepseek-v3.2	DeepSeek	$0.44	$1.76	Starter
glm-5	Zhipu	$0.44	$1.98	Starter
qwen3.5-plus	Alibaba	$0.44	$2.64	Starter
qwen3.5-flash	Alibaba	$0.13	$1.32	Starter
qwen3-max	Alibaba	$0.77	$3.08	Starter
kimi-k2.5	Moonshot	$0.44	$2.31	Starter
MiniMax-M2.5	MiniMax	$0.23	$0.92	Starter

Free / Promotional

Model ID	Input / MTok	Output / MTok	Tier
nano-banana-2	$0.05	$0.30	Starter

10. Authentication

All API keys use the ck_ prefix. The proxy accepts multiple authentication header formats:

Header	Format	Used By
x-api-key	ck_...	Claude Code, Anthropic SDK
Authorization	Bearer ck_...	OpenCode, OpenAI SDK, cURL
anthropic-api-key	ck_...	Alternative Anthropic header
x-goog-api-key	ck_...	Gemini CLI compatibility

Forwarded headers

The proxy forwards anthropic-version (defaults to 2023-06-01) and anthropic-beta headers to the upstream Anthropic API. Streaming is fully supported via Server-Sent Events (SSE).

11. Rate Limits & Errors

Rate limits

Requests per hour

1,000

Minimum credit balance

200 credits ($2.00)

Streaming timeout

300s (5 min)

Rate limit info is returned in response headers: x-ratelimit-limit, x-ratelimit-remaining, x-ratelimit-reset.

Error responses

All errors return JSON:

json

{
  "error": {
    "type": "error_type",
    "message": "Human-readable description"
  }
}

Status	Type	Cause
400	invalid_request_error	Missing required parameters
401	authentication_error	Invalid or missing API key
402	insufficient_credits	Credit balance below $2.00
403	access_denied	Model not available at your tier
429	rate_limit_error	Rate limit exceeded (1,000/hr)

12. Tiers

Tiers are determined by cumulative top-up amount. Higher tiers unlock additional models.

Starter

$19.99+ cumulative top-up

Unlocks:

Claude Sonnet 4.6
GPT-5.2 / GPT-5.3 Codex
Gemini 3.0 Pro & Flash
API key creation

Opus

$99.99+ cumulative top-up

Everything in Starter, plus:

Claude Opus 4.6
Priority queue

13. Audio Generation (Fish Audio)

Text-to-speech, voice cloning, and speech recognition powered by Fish Audio. Requires Starter tier ($19.99+).

Endpoint

POST https://zenn.engineering/api/v1/audio/generations
GET  https://zenn.engineering/api/v1/audio/generations   (list models)

Available Models

Model ID	Type	Credits	Price	Required Input
audio-tts	Text to Speech	2	$0.020	text
audio-clone	Voice Clone	2	$0.020	text + audio
audio-asr	Speech Recognition	1	$0.010	audio

Text to Speech

cURL

curl -X POST https://zenn.engineering/api/v1/audio/generations \
  -H "Authorization: Bearer ck_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "audio-tts",
    "text": "Hello, welcome to Zenn!",
    "format": "mp3"
  }'

Optional parameters: format (mp3, wav, opus), temperature, reference_id (existing voice model ID).

Voice Clone

cURL

curl -X POST https://zenn.engineering/api/v1/audio/generations \
  -H "Authorization: Bearer ck_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "audio-clone",
    "text": "Say this in the cloned voice",
    "audio": "https://example.com/reference-voice.mp3"
  }'

The audio field is a URL to the reference voice sample for zero-shot cloning.

Speech Recognition (ASR)

cURL

curl -X POST https://zenn.engineering/api/v1/audio/generations \
  -H "Authorization: Bearer ck_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "audio-asr",
    "audio": "https://example.com/speech.mp3",
    "language": "en"
  }'

Response Format

TTS / Voice Clone returns an audio URL:

TTS Response

{
  "model": "audio-tts",
  "created": 1709654321,
  "data": [{ "url": "https://storage.zenn.run/generated/audio/..." }]
}

ASR returns transcribed text:

ASR Response

{
  "model": "audio-asr",
  "created": 1709654321,
  "text": "Hello, welcome to Zenn!",
  "duration": 2.5
}

Ready to start?

One key works across Claude Code, OpenCode, OpenClaw, Codex CLI, Gemini CLI, and Cursor. Top up credits and create your API key.

Get Credits Manage API Keys