Test which OpenRouter models your key can actually access
Compare OpenRouter’s user-scoped model list with its general catalog and diagnose whether extra models come from OpenRouter, a stale BoltAI cache, or a catalog fallback.
Both requests go directly from this browser to OpenRouter with caching disabled. BoltAI never receives or stores your API key.
Direct OpenRouter check
Compare user access with the full catalog
1. Your browser
Key lives in this tab
2. Direct request
openrouter.ai only
3. Local report
Compared in memory
This tool calls both OpenRouter model endpoints directly with cache disabled. It never writes the key to cookies, browser storage, analytics events, or BoltAI servers.
Read the comparison
Pinpoint the wrong model-list path
OpenRouter documents the user endpoint as filtered by provider preferences, privacy settings, and guardrails. The general endpoint is the broad catalog. The difference is the evidence needed to diagnose BoltAI’s picker.
See OpenRouter’s models documentation.
Use a key you control
Run this only on a trusted device. If you prefer, create a temporary OpenRouter key for the diagnostic and revoke it when finished.
- 01
Check Available to key
This is the live result from/api/v1/models/user. It is the model list BoltAI should display for the same key. - 02
Check Catalog-only
These models exist in OpenRouter’s general catalog but were excluded by the key’s effective account configuration. They should not appear in BoltAI’s model manager. - 03
Compare the same key in BoltAI
Confirm BoltAI is using exactly the key tested above. Then refresh its OpenRouter model list and compare several catalog-only IDs with the model picker. - 04
Classify the failure
If BoltAI shows catalog-only IDs, suspect an app cache or a fallback to/api/v1/models. If both lists match, review the OpenRouter account settings instead. - 05
Send the safe report
Copy the sanitized report and include the BoltAI version plus whether the list changed after Refresh Model List. Never send the API key itself.
Prefer Terminal for a production key?
Ask OpenRouter for the user-scoped list without putting the key in shell history.
curl -s https://openrouter.ai/api/v1/models/user \
-H "Authorization: Bearer $OPENROUTER_API_KEY" | \
jq -r '.data[].id' | sortWhat the safe report proves
It captures a fresh user endpoint result, the full-catalog comparison, model counts, request metadata, and the exact user-available IDs without exposing the key.
That is enough to separate an OpenRouter configuration issue from BoltAI caching or endpoint fallback behavior.
Frequently asked questions
Before you paste a key
- Is it safe to enter my OpenRouter API key here?
- The test runs locally in your browser. Your browser sends the key directly to OpenRouter, and BoltAI does not receive, upload, store, log, or share it. The key is not added to cookies, analytics, or browser storage.
- Which OpenRouter endpoint lists models available to my key?
- OpenRouter documents GET /api/v1/models/user as the user-scoped model endpoint. It filters models using the account’s provider preferences, privacy settings, and guardrails. GET /api/v1/models is the general catalog.
- Why does this tool call two model endpoints?
- It calls /api/v1/models/user to get the list available under the key’s effective settings, then calls /api/v1/models to get the general catalog. Comparing the two reveals which catalog models should not appear in BoltAI for that key.
- Does the checker bypass browser or CDN caches?
- The browser requests use cache: no-store and the comparison is created in memory. This rules out this page’s browser cache, but it cannot inspect or clear a model list already cached inside the BoltAI app.
- What if the checker shows fewer models than BoltAI?
- If /models/user returns fewer models and BoltAI shows catalog-only models, the OpenRouter account is filtering correctly. BoltAI is then likely showing a stale cached result or a list obtained from the general catalog or fallback path.
- What if the checker shows the entire OpenRouter catalog?
- OpenRouter’s effective provider preferences, privacy settings, and guardrails are not excluding additional models for that key at the time of the test. Review the account settings and confirm you tested the same key used by BoltAI.
- Does this test send a prompt or spend credits?
- No. It only requests model metadata. It does not send a chat prompt, generate a response, or spend inference credits.
- What should I send to BoltAI support?
- Use Copy safe report. It includes endpoint status, model counts, the diagnosis, and the user-available model IDs. It omits the API key. Never send the raw key to support.
Use your OpenRouter models in BoltAI
Download the native Mac app and connect OpenRouter with OAuth or your own API key.