ed083ae19c
Co-authored-by: hannesrudolph <49103247+hannesrudolph@users.noreply.github.com> Co-authored-by: roboclaw-bot <309084314+roboclaw-bot@users.noreply.github.com>
5.6 KiB
summary, title, read_when
| summary | title | read_when | ||
|---|---|---|---|---|
| Use Requesty's unified OpenAI-compatible router to access hundreds of models in OpenClaw | Requesty |
|
Requesty routes requests to hundreds of models behind a single OpenAI-compatible endpoint and API key, with per-request pricing and prompt caching.
| Property | Value |
|---|---|
| Provider | requesty |
| Distribution | Bundled with OpenClaw |
| Auth | REQUESTY_API_KEY |
| API | OpenAI-compatible |
| Base URL | https://router.requesty.ai/v1/ |
| Model catalog | Auto-discovered via /v1/models/managed |
Requesty ships with OpenClaw as a bundled provider plugin. No separate plugin installation is required; it updates together with OpenClaw.
Setup
Go to [app.requesty.ai](https://app.requesty.ai), sign in or create an account, then generate an API key. ```bash openclaw onboard --auth-choice requesty-api-key ```Or set the environment variable directly:
```bash
export REQUESTY_API_KEY="<your-requesty-api-key>" # pragma: allowlist secret
```
Onboarding preserves your model entries and leaves generated catalog rows to discovery. With models.mode: "replace", it also writes the built-in catalog because that mode skips discovery.
Non-interactive example
openclaw onboard --non-interactive --accept-risk --skip-health \
--mode local \
--auth-choice requesty-api-key \
--requesty-api-key "$REQUESTY_API_KEY"
Default model and catalog
The default model is requesty/claude-sonnet-5-5.
Once a key is configured, OpenClaw queries GET https://router.requesty.ai/v1/models/managed
with that key and publishes the managed policies your key can route to, with Requesty's live
prices, context window, output limit, vision, and reasoning flags. Managed policies are
addressed by bare name (claude-sonnet-5-5, gpt-6-luna, gemini-3.8-flash), and
Requesty picks the hosting provider. When discovery is unavailable, the static catalog
keeps those three defaults addressable.
Run /models requesty or openclaw models list --provider requesty to see the list for your
key.
Config example
{
env: { vars: { REQUESTY_API_KEY: "<your-requesty-api-key>" } }, // pragma: allowlist secret
agents: {
defaults: {
model: { primary: "requesty/claude-sonnet-5-5" },
},
},
}
Behavior notes
Discovery lists Requesty's managed policies, which route a bare model name to the best available hosting provider under your organization's routing and compliance settings. To pin a specific host instead (`vertex/claude-sonnet-4-6`, `bedrock/claude-sonnet-4-6`), add that id under `models.providers.requesty.models` in your config; OpenClaw publishes authored rows alongside the discovered policies. Requests use the OpenAI-compatible chat completions path with a Bearer token built from your API key. Streaming, tool calls, and vision input work for models whose catalog flags advertise them. `/think` levels are sent as the standard `reasoning_effort` field for models that report `supports_reasoning`.Gemini-backed refs stay on the proxy-Gemini replay path: OpenClaw sanitizes Gemini
thought signatures there but does not enable native Gemini replay validation.