OpenCode on 600+ models, one /connect away
Requesty is a built-in provider in OpenCode. Run /connect, paste a key and pick from every model your organization approves. Add routing policies and models outside the catalog in opencode.json.
{ "$schema": "https://opencode.ai/config.json", "model": "requesty/claude-sonnet-4-6", "provider": { "requesty": { "options": { "baseURL": "https://router.requesty.ai/v1", "apiKey": "{env:REQUESTY_API_KEY}" } } } }
Set up OpenCode in three steps
The built-in provider already knows the base URL, so /connect only asks for a key. Use opencode.json when you want a config in git, the EU endpoint or your own routing policies.
- 01Run /connect
Open OpenCode in a project, run /connect and choose Requesty. From a shell, opencode auth login does the same.
- 02Paste your key
OpenCode stores it in ~/.local/share/opencode/auth.json. For a checked-in config, read it from REQUESTY_API_KEY in opencode.json instead.
- 03Pick a model
Run /models to see the Requesty models approved for your organization, then set a default with /model.
{ "model": "requesty/policy/coding-default", "provider": { "requesty": { "models": { "policy/coding-default": { "name": "Coding default (policy)" } } } } }
Add policies and models outside the OpenCode catalog
The built-in provider lists bare managed names such as requesty/claude-sonnet-4-5. Your own routing policies and newer models appear once you declare them in opencode.json, under provider.requesty.models or a second OpenAI-compatible provider.
Without the models entry, OpenCode answers a policy with Unexpected server error.
eventual success on a routing policy in April 2026, against 85.01% for callers pinned to one provider.
of provider failures on the gateway in April 2026 were 429s. Throttling, not downtime, is what stops an agent mid-task.
requests a day on free models, with no credit card, through the same key you connect to OpenCode.
Cost per branch, repo and developer
One install script wraps the opencode command and forwards your git branch, repo, user and OpenCode version as headers. Requesty strips them before the provider, and your dashboard breaks spend down by each.
- One policy for the teamChange the model behind it once and every session switches
- Approved models onlyThe picker follows your organization's approved list
- EU endpointSet baseURL to https://router.eu.requesty.ai/v1
OpenCode with Requesty, answered
How do I configure OpenCode with Requesty?
Run /connect inside OpenCode, choose Requesty and paste your API key. For a file-based setup, add a requesty entry under provider in opencode.json, in the project root or ~/.config/opencode/opencode.json, with baseURL set to https://router.requesty.ai/v1 and apiKey set to {env:REQUESTY_API_KEY}.
Where does OpenCode store the API key?
/connect and opencode auth login write it to ~/.local/share/opencode/auth.json. OpenCode also reads REQUESTY_API_KEY from the environment, which is the better choice for a config you commit.
How do I add a model that is missing from OpenCode's /models list?
Add a second provider in opencode.json that uses @ai-sdk/openai-compatible with baseURL set to https://router.requesty.ai/v1, and list the Requesty model IDs you want under models. They appear in /models under their own group and route through Requesty like the built-in provider.
How do I use a routing policy in OpenCode?
Declare the policy under provider.requesty.models in opencode.json, for example policy/coding-default, then reference it as requesty/policy/coding-default. Change the models behind the policy in the Requesty dashboard and every session picks it up without a config edit.
Are there free models for OpenCode?
Requesty gives 200 requests a day on free models, with no credit card, through the same key you connect to OpenCode. Paid models are billed per token at the provider price plus 5%.
How do I track token usage and cost in OpenCode?
Every request shows up in the Requesty analytics dashboard with tokens, cost and latency. Install the wrapper with curl -fsSL https://www.requesty.ai/opencode/install.sh | bash to add cost per git branch, repo and developer.
How do I change the model in OpenCode?
Run /models and pick one for the session, or set model in opencode.json for a default. Point the default at a routing policy if you expect to change it often.
How do I fix rate limit errors in OpenCode?
A rate limit error means the provider is throttling you. Use a Requesty fallback policy as your model and a throttled request retries on the next model in the chain. Requesty itself does not limit requests per minute.
Can OpenCode use the EU endpoint?
Yes. Set baseURL to https://router.eu.requesty.ai/v1 under provider.requesty.options in opencode.json and OpenCode traffic goes through the Frankfurt gateway.
Guides for the other agents you run
The same key and the same dashboard work across every harness, so spend and logs stay in one place.
Run OpenCode on Requesty
600+ models, a spend cap on every key, and every session on the record.