01 / START
Quick start
- Sign up or sign in with Rate Cloud. Confirm your email. For the free three-hour trial, link Telegram in the dashboard; after the trial ends, a subscription to the selected model is required.
- Open Settings → API keys and create a key. Copy the secret right away: once you close the window it is never shown again.
- Open Connecting in the dashboard, select the app and the operating system. The “Automatically” tab shows a ready-made command with your key if automatic installation is supported.
- Paste the command into Terminal or PowerShell, wait for the success message, restart the program, and select Rate Cloud. If you prefer manual setup, open the «Manual» tab on the same page.
- Send a short test message. The model list in the app shows all five entries; a request to a model will succeed with an active subscription or trial access and suitable key access.
Automatic setup changes only the configuration of the selected program and saves a backup of the existing file. Chats are not deleted. The disconnect command is located next to the connect command in the dashboard.
02 / ACCESS
How to create an API key
- In the dashboard, click «Settings», scroll to the «API keys» section, and click «+ Create key».
- Give the key a clear name, such as «Codex on laptop». The name is only for your reference.
- In «Model Access», select «All active models» if you want to automatically use future subscriptions. The «Selected models only» option restricts the key to the listed models; purchasing another subscription will not lift this restriction by itself.
- Select IP access: «Any IP» for a regular laptop, or a list of allowed addresses for a permanent server. Set the expiration period. If you use Codex or Claude Code, enable the «Codex / Claude Code support» checkbox.
- Click «Create key» and copy the displayed secret. Store it like a password. If you lose the secret, create a new key. Do not paste the key into a public repository or a support request.
The key controls access; the subscription grants the right to make requests to a specific model. The list of all five models may be displayed even without a subscription; this avoids having to reconfigure the client after purchase. Codex and Claude Code require a key with compatibility enabled. Open API keys →
03 / CATALOG
All five models
In the field model When calling the API, pass the ID exactly as shown in the table. The ID is written in Latin characters, lowercase, with dots and hyphens exactly as in the table. Do not add a prefix to the ID ratecloud/ in the direct API: it is only needed to select a provider inside some apps.
| Name | ID for HTTP API | Selection in clients |
|---|---|---|
| GLM 5.3 | glm-5.3 | Rate Cloud / GLM 5.3 |
| GLM 5.3 Flash | glm-5.3-flash | Rate Cloud / GLM 5.3 Flash |
| DeepSeek V4.1 Flash | deepseek-v4.1-flash | Rate Cloud / DeepSeek V4.1 Flash |
| DeepSeek V4 Pro | deepseek-v4-pro | Rate Cloud / DeepSeek V4 Pro |
| Qwen 3.8 Max | qwen-3.8-max | Rate Cloud / Qwen 3.8 Max |
Exact names for Codex and Claude Code
All options are listed here in full. In Codex, specify the desired line in model, in Claude Code — in ANTHROPIC_MODEL.
| Model | Codex | Claude Code |
|---|---|---|
| GLM 5.3 | ratecloud/glm-5.3 | ocx-claude-ratecloud--glm-5.3 |
| GLM 5.3 Flash | ratecloud/glm-5.3-flash | ocx-claude-ratecloud--glm-5.3-flash |
| DeepSeek V4.1 Flash | ratecloud/deepseek-v4.1-flash | ocx-claude-ratecloud--deepseek-v4.1-flash |
| DeepSeek V4 Pro | ratecloud/deepseek-v4-pro | ocx-claude-ratecloud--deepseek-v4-pro |
| Qwen 3.8 Max | ratecloud/qwen-3.8-max | ocx-claude-ratecloud--qwen-3.8-max |
A subscription for each model is purchased separately on the the pricing page. Check trial access and model availability in the dashboard. If you are subscribed to multiple models, the automatic command defaults to the model with the oldest active subscription. In Codex clients, the IDs look like ratecloud/glm-5.3, and in Claude Code — ocx-claude-ratecloud--glm-5.3; in the same way, substitute any of the other four IDs.
04 / DEVELOPERS
HTTP API and your first request
| OpenAI-compatible Base URL | https://llm.rate-cloud.com/v1 |
|---|---|
| Model catalog | GET https://llm.rate-cloud.com/v1/models |
| Chat | POST https://llm.rate-cloud.com/v1/chat/completions |
| Responses | POST https://llm.rate-cloud.com/v1/responses |
| Codex Base URL | https://llm.rate-cloud.com/codex/v1 |
| Claude Code Base URL | https://llm.rate-cloud.com/claude |
Requests to /v1 send the header Authorization: Bearer YOUR_API_KEY. First check the catalog, then send a simple request. These examples use GLM 5.3; replace the ID with one of the five models that has active access.
macOS / Linux: Terminal
printf 'Вставьте API key: '; read -r -s RATE_CLOUD_API_KEY; echo
curl -fsS https://llm.rate-cloud.com/v1/models \
-H "Authorization: Bearer $RATE_CLOUD_API_KEY"
curl -fsS https://llm.rate-cloud.com/v1/chat/completions \
-H "Authorization: Bearer $RATE_CLOUD_API_KEY" \
-H 'Content-Type: application/json' \
-d '{"model":"glm-5.3","messages":[{"role":"user","content":"Ответь: работает"}]}'
unset RATE_CLOUD_API_KEYWindows: PowerShell
$env:RATE_CLOUD_API_KEY = Read-Host 'Paste your API key'
$headers = @{ Authorization = "Bearer $env:RATE_CLOUD_API_KEY" }
Invoke-RestMethod 'https://llm.rate-cloud.com/v1/models' -Headers $headers
$body = @{ model = 'glm-5.3'; messages = @(@{ role = 'user'; content = 'Ответь: работает' }) } | ConvertTo-Json -Depth 5
Invoke-RestMethod 'https://llm.rate-cloud.com/v1/chat/completions' -Method Post -Headers $headers -ContentType 'application/json' -Body $body
Remove-Item Env:RATE_CLOUD_API_KEYFor OpenAI-compatible SDKs, set the Base URL https://llm.rate-cloud.com/v1, your key and one of the IDs from the table. For a direct Responses request, use the same key and /v1/responses.
05 / PROGRAMS
Codex CLI, app, and VS Code
Simple path: in dashboard select “Codex”, your OS, and “Automatically”. The command uses a key with “Codex / Claude Code support” enabled. After installation, open the model list and select Rate Cloud. The VS Code Codex extension uses the Codex settings.
Manual setup
- Install Codex per OpenAI instructions. Close Codex and VS Code.
- Open
~/.codex/config.tomlon macOS/Linux or%USERPROFILE%\.codex\config.tomlon Windows. If the file exists, save a copy. If not, create a text file with that name. - Add global parameters at the beginning of the file, before the sections in square brackets. Add the provider section at the end. Keep the rest of the settings.
model = "ratecloud/glm-5.3"
model_provider = "ratecloud"
[model_providers.ratecloud]
name = "Rate Cloud"
base_url = "https://llm.rate-cloud.com/codex/v1"
wire_api = "responses"
env_key = "RATE_CLOUD_API_KEY"In the terminal where you run Codex, set RATE_CLOUD_API_KEY with your secret: on macOS/Linux — export RATE_CLOUD_API_KEY='YOUR_API_KEY', in PowerShell — $env:RATE_CLOUD_API_KEY='YOUR_API_KEY'. Then, in the same window, run codex. When launching VS Code from the terminal, use code ., so the extension gets the same variable. For the other models, replace the part after ratecloud/ with the required ID from the table.
For an exact manual example for your selected OS, use the «Manual» tab in the dashboard. Check the provider format against the Codex documentation.
06 / PROGRAMS
Claude Code
Claude Code requires a key with Codex / Claude Code compatibility enabled and active access to the selected model. In dashboard select Claude Code and your OS: the automatic command will set up detection of all five models. The model name in this client has the prefix ocx-claude-ratecloud--.
Manual run without editing files
Install Claude Code, open Terminal or PowerShell and set three variables before the command claude. Replace only YOUR_API_KEY and, optionally, the model suffix.
macOS / Linux
export ANTHROPIC_BASE_URL='https://llm.rate-cloud.com/claude'
export ANTHROPIC_AUTH_TOKEN='YOUR_API_KEY'
export ANTHROPIC_MODEL='ocx-claude-ratecloud--glm-5.3'
claudeWindows PowerShell
$env:ANTHROPIC_BASE_URL='https://llm.rate-cloud.com/claude'
$env:ANTHROPIC_AUTH_TOKEN='YOUR_API_KEY'
$env:ANTHROPIC_MODEL='ocx-claude-ratecloud--glm-5.3'
claudeThese variables apply in the current terminal. Automatic installation saves the settings in ~/.claude/settings.json or %USERPROFILE%\.claude\settings.json, without affecting chat history. More about variables and settings — in the Claude Code documentation.
07 / PROGRAMS
OpenCode
The automatic command in the dashboard adds the Rate Cloud provider and all five models to the OpenCode configuration. It accounts for the format of the installed OpenCode version. On Windows, the configuration is located at %USERPROFILE%\.config\opencode\opencode.json; on macOS/Linux — in ~/.config/opencode/opencode.json.
Manual example for OpenCode v1
Create a copy of the existing file, open it in a text editor, and add the provider to the object provider. For a new file, use the full example below. Replace YOUR_API_KEY with the secret, save the file, restart OpenCode, and open /models.
{
"$schema": "https://opencode.ai/config.json",
"model": "ratecloud/glm-5.3",
"provider": {
"ratecloud": {
"npm": "@ai-sdk/openai-compatible",
"name": "Rate Cloud",
"options": {
"baseURL": "https://llm.rate-cloud.com/v1",
"apiKey": "YOUR_API_KEY"
},
"models": {
"glm-5.3": {"name": "GLM 5.3"},
"glm-5.3-flash": {"name": "GLM 5.3 Flash"},
"deepseek-v4.1-flash": {"name": "DeepSeek V4.1 Flash"},
"deepseek-v4-pro": {"name": "DeepSeek V4 Pro"},
"qwen-3.8-max": {"name": "Qwen 3.8 Max"}
}
}
}
}OpenCode v2 uses a different format providers. Do not paste the v1 example into the v2 config: a ready-made example for both versions is in the “Manual” tab in the dashboard. Official OpenCode documentation.
08 / PROGRAMS
OpenClaw, Continue, Hermes, and others
For OpenClaw, Continue, and Hermes, in dashboard there are automatic commands for Windows, macOS, and Linux, plus manual examples with file paths and copy buttons. Select the app and OS before copying the command.
| App | Where Rate Cloud is added | What to select after installation |
|---|---|---|
| OpenClaw | ~/.openclaw/openclaw.json, Windows: %USERPROFILE%\.openclaw\openclaw.json | ratecloud/glm-5.3 or another model from the catalog |
| Continue for VS Code | ~/.continue/config.yaml, Windows: %USERPROFILE%\.continue\config.yaml | The model named «Rate Cloud · …» |
| Hermes | ~/.hermes/config.yaml and the adjacent .env | Provider ratecloud and the required ID |
| Another OpenAI-compatible client | Custom provider settings in the program | Base URL https://llm.rate-cloud.com/v1, the key and the model ID |
OpenClaw manually
- Close OpenClaw. Open the file path shown above
openclaw.jsonand save the copy. - Add a provider
ratecloudвmodels.providers, without deleting other entries. Inagents.defaults.model.primaryspecify the requiredratecloud/ID. - Replace
YOUR_API_KEY, save the file and restart OpenClaw. Below is a minimal example for a new file; for an existing file, move only the Rate Cloud fields.
{
"models": {"providers": {"ratecloud": {
"baseUrl": "https://llm.rate-cloud.com/v1",
"apiKey": "YOUR_API_KEY",
"api": "openai-completions",
"models": [
{"id":"glm-5.3","name":"GLM 5.3"},
{"id":"glm-5.3-flash","name":"GLM 5.3 Flash"},
{"id":"deepseek-v4.1-flash","name":"DeepSeek V4.1 Flash"},
{"id":"deepseek-v4-pro","name":"DeepSeek V4 Pro"},
{"id":"qwen-3.8-max","name":"Qwen 3.8 Max"}
]
}}},
"agents": {"defaults": {"model": {"primary": "ratecloud/glm-5.3"}}}
}
Continue for VS Code manually
- Install the Continue extension, close VS Code, and save the copy
~/.continue/config.yaml(Windows:%USERPROFILE%\.continue\config.yaml). - To the array
models:add an entry for each model. For a new file, use the example below; for an existing file, keep its other entries and add four more blocks following the same pattern, changingnameиmodelaccording to the table. - Replace
YOUR_API_KEY, save the file and restart VS Code. In the Continue model menu, select “Rate Cloud · GLM 5.3”.
name: Rate Cloud
version: 1.0.0
schema: v1
models:
- name: Rate Cloud · GLM 5.3
provider: openai
model: glm-5.3
apiBase: https://llm.rate-cloud.com/v1
apiKey: YOUR_API_KEY
useResponsesApi: false
roles: [chat, edit, apply]To add all five models without repeating the block manually, use the automatic dashboard command: it will add the full catalog.
Hermes manually
- Close Hermes, open
~/.hermes/config.yaml(Windows:%USERPROFILE%\.hermes\config.yaml) and save a copy. - Add the provider and model as shown below. If the file already contains other settings, merge the entries without replacing the entire file.
- To the adjacent file
.envadd the lineRATE_CLOUD_API_KEY="YOUR_API_KEY". Save both files and restart Hermes.
model:
provider: ratecloud
default: glm-5.3
base_url: https://llm.rate-cloud.com/v1
api_mode: chat_completions
providers:
ratecloud:
api: https://llm.rate-cloud.com/v1
api_key: ${RATE_CLOUD_API_KEY}
models: [glm-5.3, glm-5.3-flash, deepseek-v4.1-flash, deepseek-v4-pro, qwen-3.8-max]
Zed, JetBrains AI Assistant, and other editors
- In your editor settings, find AI, Models, or Providers and add a custom OpenAI-compatible provider. If the editor does not offer a custom Base URL, this connection method is not suitable for it.
- Enter the name
Rate Cloud, Base URLhttps://llm.rate-cloud.com/v1and the API key secret. Do not add/chat/completionsin the Base URL field. - Add the model with the exact ID from the table above and select it for chat. Repeat for the remaining four if the editor does not load the list via
/models. For Zed and JetBrains, open the step-by-step «Manual» tab in the dashboard — the field names differ.
For Zed, JetBrains AI Assistant, and VS Code extensions, the dashboard shows manual steps with fields specific to the selected client. The automatic command is offered only where the installer can safely modify the configuration. Cursor does not provide a confirmed way to set a custom compatible Base URL for this scheme; do not enter your key into an unverified connection method.
09 / MANAGEMENT
How to disconnect Rate Cloud
- Open Connecting in the dashboard and select the same app and OS as during installation.
- Copy the “Disconnect” command and run it in Terminal or PowerShell. The installer will check whether the file changed after installation, then restore the previous settings from its backup. Chats and projects stay in place.
- Restart the app. If you edited the config manually after installation and the restore refuses to overwrite it, save a new copy of the file and remove only the settings
ratecloudmanually. - If the key is no longer needed, separately disable or delete it in API key settings. Removing the provider from the app does not by itself revoke the key.
10 / HELP
If something is not working
The app sees five models, but the request is rejected
The model list shows the entire catalog. Check the active subscription or trial access for the selected model, the key access settings, and its expiration date. The dashboard has a connection test with a live request.
Codex or Claude Code reports a refusal
Check the “Codex / Claude Code support” checkbox for the key and the matching endpoint: /codex/v1 for Codex, /claude for Claude Code. A regular key without this permission will not work.
The key is not shown in full
The full secret is shown at creation time. Create a new key and save it; the mask of an existing key prevents recovering the secret.
The command does not run
On Windows, use PowerShell; on macOS/Linux, use Terminal. Automatic setup requires Python 3 and access to the website. On Windows, when installing Python, check «Add python.exe to PATH», then open a new PowerShell. Do not run the Windows command in cmd.exe.
Need help with a specific program
On the connection page select the app and OS: it shows the exact commands, paths, and manual fields. When contacting support state the app name, OS, and the error text, but do not send the key secret.
