Claude by Anthropic
Native Anthropic Messages API, ready for Claude Code.
- Claude Opus 5.5Frontierclaude-opus-5.5 ยท 262K ctx$5.40 in
$36.00 out - Claude Fable 5.2Frontierclaude-fable-5-2 ยท 262K ctx$5.40 in
$36.00 out
One API key for every Claude, ChatGPT and Gemini model, and DeepSeek, Grok, Kimi and more. Plug it into Claude Desktop, Claude Code, Codex, Cursor, Roblox Studio MCP, Blender MCP or your own code, and switch between 256 models without rewriting a line.
An animation: Claude Code connects with an Evolved key on Claude Opus 5.5 and builds an obby in Roblox Studio, then models a low-poly rocket in Blender and renders it. Next, the Evolved dashboard's chat asks Claude Opus 5.5, Prompt Creator makes a prompt, and the chat switches to GPT-6 Astra.
API keys sold
To developers and teams worldwide, with more every day.
Think of it as Claude Max and ChatGPT Pro in one, with Gemini, DeepSeek, Grok, Kimi and the rest of our 256 models from 16 families on the same balance.See the note about ChatGPT Pro, Claude Max and Google AI Ultra
Paste your key into Claude Desktop once, then ask Claude to work on the files on your computer.
Claude, ChatGPT and more
Frontier reasoning, balanced everyday models and fast lightweight ones, all on the same key, the same balance and one rate table.
Native Anthropic Messages API, ready for Claude Code.
Native Responses and Chat Completions, ready for Codex.
Gemini, DeepSeek, Grok, Kimi, Qwen, Mistral, Llama, GLM, MiniMax, MiMo, Nemotron, Sonar, Command and Hunyuan too: 256 models on the same key.
More modelsPricing
Pick Claude, ChatGPT, Gemini or a mix. Credit packs from $20, worth up to 16 times what you pay, on one balance for every model in your plan. No subscription.
Claude All Models
Fable 5.2 & Opus 5.5NewThink of it as Claude Max on a prepaid balance.*
$50 credit
$20one-time
Try every model on a real project.
$100 credit
$30one-time
Daily coding sessions and side projects.
$200 credit
$40one-time
For professionals shipping with agents every day.
$400 credit
$55one-time
Heavy agent workloads, all day long.
$1,000 credit
$95one-time
Teams and long-running automations.
$2,000 credit
$150one-time
The most credit for every dollar.
Buying for a team, or need more than $2,000 in credit?
Write to our sales team and we will set up the credit and keys that fit.
* ChatGPT Pro, Claude Max and Google AI Ultra are subscription plans sold by OpenAI, Anthropic and Google. evolved.to is independent and does not sell those plans; your credit pays for API access to the same Claude, ChatGPT and Gemini model families.
Why evolved.to
Keep your tools, your SDKs and your prompts. Change one base URL and every model is on the table.
A single key replaces separate provider accounts, keys and invoices. Manage it, add credit and revoke it from one dashboard.
Plan with a frontier model, build with a balanced one, fan out sub-agents on a fast one. Just by name.
Anthropic and OpenAI wire formats, served as-is. Official SDKs and clients work without shims.
Swap to a newer or cheaper model without touching your integration. It is a one-line change.
client.messages.create({- model: "claude-sonnet-4.5",+ model: "claude-sonnet-5", max_tokens: 16000, })
See your balance, token usage and request history per key, and add credit in seconds.
How it works
No new SDK, no proxy to run, no prompt rewrites.
Buy a credit pack and your key is ready in the dashboard. It is shown once, in full, so store it like a password.
Set the base URL to api.evolved.to and use your key. Claude Code, Codex, Cursor and the SDKs all support it.
Send claude-opus-5, gpt-5 or any other model ID. Switch whenever you like; your integration stays the same.
export ANTHROPIC_BASE_URL="https://api.evolved.to"
export ANTHROPIC_AUTH_TOKEN="$EVOLVED_API_KEY"
export ANTHROPIC_MODEL="claude-opus-5"
claude# ~/.codex/config.toml
model = "gpt-5"
model_provider = "evolved"
[model_providers.evolved]
name = "evolved.to"
base_url = "https://api.evolved.to/v1"
env_key = "EVOLVED_API_KEY"
wire_api = "responses"import os
import anthropic
client = anthropic.Anthropic(
base_url="https://api.evolved.to",
api_key=os.environ["EVOLVED_API_KEY"],
)
message = client.messages.create(
model="claude-opus-5",
max_tokens=16000,
messages=[{"role": "user", "content": "Hello, Claude"}],
)
for block in message.content:
if block.type == "text":
print(block.text)import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.evolved.to/v1",
apiKey: process.env.EVOLVED_API_KEY,
});
const response = await client.responses.create({
model: "gpt-5",
input: "Hello!",
});
console.log(response.output_text);Compatible clients
Terminal agents, editors, desktop apps, automations, games and SDKs. Every guide takes a couple of minutes.
Run every Claude model in your terminal with two environment variables.
Use the Claude Code extension inside your editor with the same key.
Chat with every Claude model in the Claude Desktop app, on your Evolved key.
Add evolved.to as a model provider in Codex and use GPT models.
Override the OpenAI base URL and add any model by name.
Pick the Anthropic or OpenAI Compatible provider and paste your key.
Configure an OpenAI Compatible profile in a few clicks.
Use any model in VS Code's chat, or through an extension.
Point the Hermes Agent CLI at a custom endpoint.
Add evolved.to as a provider for your personal AI agent.
Let the desktop agent work on your files with any model.
Bring your own key to Factory's coding agent.
Use any model in your workflows through an OpenAI credential.
Build your game with Claude Code, Cursor or Codex through Studio's MCP server, or give NPCs an AI.
Model, light and render by asking, through the Blender MCP add-on.
Every MCP server, local or remote, in every app that speaks MCP.
Official anthropic and openai packages with a custom base URL.
Security
Your requests, your keys and your usage data are handled on a least-exposure principle, end to end over TLS 1.3.
Keys are hashed before they are stored and shown in full only once. Revoke or rotate one in a click, without touching the others.
Every request is validated (key, balance and rate limits) before it reaches a model. Inactive keys are rejected at the edge.
Request content is kept encrypted for at most 7 days for support, then deleted. Full keys and authorization headers are never logged.
FAQ
Something else? Our docs cover every client and error code, and support answers from your dashboard.
Get a key, set a base URL, and ship with every Claude, ChatGPT and Gemini model today.
Open your dashboard to check your balance, token usage and request history, or add credit in seconds.