50+ models · multiple providers · one API key

One account.Every top AI.

Claude, GPT, Gemini, Grok and a dozen more — in the chat, or through one API key wired straight into your own code.

✓Compliant✓Stable✓Private✓Simple

Claude Code · drop-in gateway

export ANTHROPIC_BASE_URL=https://kovelo.ai/api
export ANTHROPIC_API_KEY=kovelo-••••••••••••
claude   # every top model, one key, official APIs

Every major model, one place

Anthropic

OpenAI

Gemini

xAI

DeepSeek

Qwen

Kimi

MiniMax

Built for the way you actually work

Chat with any model, or wire one key into your own apps.

Claude Code gateway

One API key, every model, dropped straight into your own code and tools.

Multi-model chat

One thread, any model. Switch mid-conversation with scene presets.

AI image studio

Crisp, accurate images from GPT-Image-2; fast, expressive style from Grok Imagine.

Real-time search

Live web answers when a question needs today's information.

Privacy first

Encrypted in transit, strict per-account data isolation, never shared with third parties.

Flexible billing

Pay as you go or subscribe. Credits never expire.

Built for agents & automation

Point any agent, bot, or fleet at Kovelo. Lock or switch each key's model from your dashboard — no config edits, no downtime. Works with Claude Code, OpenClaw, Cline, Cursor, and any tool that speaks the OpenAI or Anthropic API.

Works with your favorite agents and tools

Codex

Claude CodeCCursor

OpenClaw+Any OpenAI / Anthropic-compatible tool

For developers

Claude Code, OpenClaw, and every agent — one key, every model

✓Drop-in base URL, zero code changes
✓Lock a model per key — switch from your dashboard, no downtime
✓Works with Claude Code, OpenClaw, Cline, Cursor, and more

For developers

Run OpenAI Codex on Kovelo models

Both the Codex CLI and desktop app work with Kovelo. Add three lines of config and Codex reads your code, edits files, runs tests and ships — powered by Kovelo models.

✓CLI and desktop app, one config for both
✓Switch models right inside Codex — no config edits, no restart
✓Full agent capability: read and write files, run commands, run tests
✓Usage bills to the same key and the same balance as everything else

Live: switching between Kovelo-provided models inside Codex

AI Coding

You don't need to write code to build your own product

You have a product idea, but you can't code and can't afford a technical co-founder — that used to be the end of the road. Not anymore. Here, top models take you from an idea to something buildable, step by step, then hand it off to Claude Code or Cursor to write the actual code.

✓From a rough idea to a spec an engineer can build from
✓See the UI mockup before you commit to building it
✓Hand off to Claude Code or Cursor to write the actual code

From an idea to scoping, spec, UI mockup, and finally handing off to code

Scope it

Is this worth building? Where does v1 stop? Who's it for first?

Write the PRD

Turn the idea into a spec an engineer can build from

Mock the UI

Generate interface mockups so you can see it before it's built

Write the build brief

A clearly scoped task list, ready for Claude Code

Your job is knowing what you want. Writing the code isn't the hard part anymore.

Batch Mode

Half price for work that can wait

Submit work in bulk, get results within 24 hours, at half the price of realtime calls. Not everything needs an instant answer.

✓Rewriting product descriptions and marketing copy in bulk
✓Classifying and tagging support tickets and user feedback
✓Summarising documents, cleaning and structuring data
✓Model evaluations and regression testing

Built for: Teams with backlogs of data, AI products running scheduled evals, and content-heavy commerce and media.

✓Up to 50,000 requests per job
✓Failed rows aren't billed — you only pay for what succeeds
✓Compatible with the OpenAI Batch format — little to no code change
✓Works with your existing API key

Four steps

1Prepare a JSONL file (one request per line)

{"custom_id":"req-1","method":"POST","url":"/v1/responses","body":{"model":"<模型ID>","input":"你的提示词","max_output_tokens":500}}

2Upload

curl -X POST https://kovelo.dev/v1/files \
  -H "Authorization: Bearer $KOVELO_API_KEY" \
  -F "purpose=batch" -F "file=@input.jsonl"

3Create the batch job

curl -X POST https://kovelo.dev/v1/batches \
  -H "Authorization: Bearer $KOVELO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input_file_id":"file-kv_xxx","endpoint":"/v1/responses","completion_window":"24h"}'

4Retrieve the output

curl https://kovelo.dev/v1/files/file-kv_xxx/content \
  -H "Authorization: Bearer $KOVELO_API_KEY" > output.jsonl

Grok Imagine

GPT-Image-2

Image Studio

Two image engines, one creative canvas

Generate with GPT-Image-2 for crisp, accurate composition, or Grok Imagine for fast, expressive style — switch in a click.

✓GPT-Image-2 · precise prompt control
✓Grok Imagine · fast, stylized output
✓Transparent pricing

Multiple models review one document in turn and produce a finished draft

Multi-Model Relay

Important documents deserve more than one set of eyes

When a single model checks its own work, it just polishes within its own blind spots — the prose reads better, but the problems are still there. Multi-Model Relay lines up 2 to 5 models from different vendors: the first pass finds the problems, the middle passes rewrite, and the last hands you a finished document.

✓Cross-vendor — only a different vendor catches what one missed
✓A relay, not a vote: each pass builds on everything before it
✓Ships a finished file — a downloadable Word doc, not feedback

Available in Documents, Code, and Write.

Accepted payments

Visa

Mastercard

Amex

Alipay

WeChat Pay

Apple Pay

Google Pay

Link

Why Kovelo

We compete on stability and compliance — not gray-market price.

🏛️

Real US company

Official vendor APIs, proper invoices, no overnight shutdowns.

🔀

Never locked into one vendor

50+ models from the major providers. If one goes down, rate-limits you, or tightens its policy overnight, switch to another and keep going — no code changes, no waiting for it to come back.

🔐

Yours alone

Local encryption; we keep only anonymous behavior metadata.

Choose your plan

Free

For trying it out

✓Access to all 30+ top AI models
✓1 API key · 20 RPM
✓A first taste of chatting with the world's best AI brains
✓Basic fallback, then stops after 48 hours

Starter Lite

For light personal use

$3.99/mo

✓All 50+ models, one account
✓5 API keys · 60 RPM
✓Claude Code gateway, lock a model per key
✓Pay-as-you-go, auto-deduct balance over the cap

Starter

For regular individual users

$9.99/mo

✓All 30+ models, switch anytime
✓10 API keys · 120 RPM
✓Claude Code gateway, lock a model per key
✓Real-time web search across models
✓Pay-as-you-go, auto-deduct balance over the cap

Start free today

No commitment. Upgrade only when you're ready.

About Kovelo

The world's best AI minds shouldn't live behind a dozen separate accounts, scattered keys, and walled gardens. Kovelo brings them together — Claude, GPT, Gemini, Grok, and 50+ leading models — into one account, one API key, one place to build.

We believe access to frontier intelligence should be simple, stable, and trustworthy. So we connect directly through each vendor's official API endpoints — never resold or gray-market channels. Account data is strictly isolated and never shared with third parties. And we compete on reliability and compliance — not on cutting corners.

Kovelo LLC is a California-registered company built for developers, teams, and creators around the world — the people turning today's most powerful AI into real products, real work, and real ideas.

One account. Every top AI. Built to last.

Contact & support

Email support

support@kovelo.ai

@kovelo_support

WeChat (China users only)