Tool selection guide
Compare Cursor, Continue, Open WebUI, and LiteLLM paths before you route real developer traffic. TKEN fits at the model API layer: tools call an OpenAI-compatible base URL, and TKEN receives those requests at https://www.tken.shop/v1.
Developer tool, chat UI, proxy, or internal app
-> OpenAI-compatible base URL
-> TKEN https://www.tken.shop/v1
Use a low-limit key for first rollout.
Verify /v1/models before team traffic.Pick the integration point by the control boundary you need: personal IDE testing, direct chat UI setup, shared proxy routing, or per-user virtual keys.
Best when your installed Cursor settings expose a compatible custom provider path for the workflow you are testing. Cursor documents custom API-key limits, so verify the current UI before planning a team rollout. If the needed base URL path is not available, move TKEN into Continue, Open WebUI, LiteLLM, SDK code, or another tool that explicitly supports custom OpenAI-compatible endpoints.
Best for one developer or a small workspace that wants an IDE assistant with explicit OpenAI-compatible apiBase configuration. Keep the key in a secret reference or environment variable, test one low-risk prompt first, and keep file-write or agent behavior behind human approval during evaluation.
Best for a fast personal or low-volume browser chat UI. Point the OpenAI-compatible provider URL at https://www.tken.shop/v1, keep the TKEN key server-side, and verify model discovery before shared use.
Best when a team wants a chat UI plus central route names, proxy auth, logs, retries, fallback behavior, and spend controls. Open WebUI calls LiteLLM; LiteLLM owns the upstream TKEN key and route map.
Best for internal apps, agents, chat UIs, and individual users that need separate keys, budgets, model allowlists, RPM/TPM limits, expiration windows, and owner metadata without sharing the upstream TKEN key.
Run the same endpoint checks before configuring a UI, IDE, proxy, or agent. This separates key, account, and model-route problems from tool-specific settings.
export TKEN_API_KEY="sk-your-tken-key"
export TKEN_BASE_URL="https://www.tken.shop/v1"
export TKEN_MODEL="replace-with-an-available-model"
curl "$TKEN_BASE_URL/models" \
-H "Authorization: Bearer $TKEN_API_KEY"
curl "$TKEN_BASE_URL/chat/completions" \
-H "Authorization: Bearer $TKEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "'"$TKEN_MODEL"'",
"messages": [
{"role": "user", "content": "Reply with one short sentence."}
]
}'Confirm /v1/models returns the route you plan to use, then record the exact model ID in the tool, UI, or proxy config.
Start with a low-limit key, a single model route, and a small prompt before exposing the setup to teammates or agent loops.
Decide how 401, 404, 429, timeouts, repeated 5xx responses, and unexpected spend stop the rollout before adding more tools.
Do not put API keys, account IDs, private prompts, source code, usage records, screenshots with secrets, or support transcripts into public repos or community posts.
Use these pages after choosing a path. The UTM links keep this tool-selection page tied to the v0.15 developer release.
Use a terminal preflight with TKEN_API_KEY, TKEN_BASE_URL, and one available model ID before editing any IDE, UI, proxy, or agent config.
Add LiteLLM when you need stable route names, virtual keys, per-user budgets, team-level limits, fallback behavior, or one proxy endpoint shared by multiple tools.
Do not assume that. Cursor documents custom API-key behavior with limits, so verify the installed settings and use TKEN in explicit OpenAI-compatible endpoint workflows where the base URL is supported.
No. TKEN is an independent third-party OpenAI-compatible API gateway and is not officially affiliated with Cursor, Continue, Open WebUI, LiteLLM, OpenAI, or other model providers.
Create a TKEN key, test https://www.tken.shop/v1 from the terminal, then choose the lowest-control path that matches your workflow before adding team traffic.