Releases: vivian254338489/tken-openai-compatible-examples
v0.25.0 Integration decision map
v0.25.0 adds an integration decision map for developers choosing the first TKEN or OpenAI-compatible gateway rollout path:
- 5-minute path picker across API clients, OpenAI SDK apps, Vercel AI SDK, LangChain, LlamaIndex, Open WebUI, LiteLLM, coding tools, CI, agents, and MCP hosts
- default sequence for /models, non-streaming chat, redacted smoke evidence, and optional capability tests
- risk-signal table for endpoint, model, streaming, tool-call, UI, and agent cost failures
- adoption paths for existing SDK apps, Open WebUI teams, and agent or MCP hosts
- README and docs navigation updates so new visitors can choose one first integration instead of wiring every tool at once
Disclosure: TKEN is an independent third-party OpenAI-compatible API gateway and is not officially affiliated with OpenAI or other model providers or tools.
v0.24.0 OpenAI SDK base URL quickcheck
v0.24.0 adds an OpenAI SDK base URL quickcheck pack for developers migrating an existing OpenAI SDK app to TKEN:
- no-network CLI for validating /v1 base URL shape before live endpoint tests
- API-key environment presence check without printing key values
- model-selection reminder before the first SDK chat request
- JSON output for CI or local migration notes
- guide covering Node.js �aseURL, Python �ase_url, migration sequence, compatibility matrix, and common failure causes
- README, docs navigation, tools documentation, package metadata, and release checks for SDK base URL migration intent
Disclosure: TKEN is an independent third-party OpenAI-compatible API gateway and is not officially affiliated with OpenAI or other model providers.
v0.23.0 Smoke evidence runner
v0.23.0 adds a smoke evidence runner for developers who need safe, paste-ready endpoint validation artifacts before filing issues or rollout notes:
- no-dependency CLI for Markdown or JSON evidence after GET /models and optional non-streaming POST /chat/completions
- redacted status, latency, model discovery, selected model, failure class, and response-shape evidence
- no API keys, bearer tokens, private prompt text, full response text, or provider account data stored in evidence output
- guide for issue reports, release checklists, team handoffs, and rollout records
- README, docs navigation, tools documentation, package metadata, and release checks for redacted evidence capture
Disclosure: TKEN is an independent third-party OpenAI-compatible API gateway and is not officially affiliated with OpenAI or other model providers.
v0.22.0 Postman and Bruno API client smoke tests
v0.22.0 adds importable API-client smoke-test collections for developers validating TKEN before SDK, UI, proxy, or agent integration:
- Postman collection and local environment for GET /models and POST /chat/completions
- Bruno collection with local .bru files and TKEN_API_KEY environment-variable handling
- guide for variables, import flow, safe redacted evidence, and common endpoint failures
- README, docs navigation, tool integration notes, tools documentation, package metadata, and release checks for no-code endpoint validation
Disclosure: TKEN is an independent third-party API gateway and is not officially affiliated with Postman, Bruno, OpenAI, or other model providers.
v0.21.0 Open WebUI + LiteLLM Docker Compose stack
v0.21.0 adds a Docker Compose pack for developers starting Open WebUI and LiteLLM together while routing LiteLLM to TKEN:
- Compose stack for Open WebUI -> LiteLLM -> TKEN
- LiteLLM route config that maps tken-chat to a TKEN OpenAI-compatible upstream model
- environment template that keeps the upstream TKEN key in LiteLLM and gives Open WebUI a LiteLLM key
- preflight checks for /v1/models and one non-streaming chat request before connecting UI users
- README, docs navigation, tool integration notes, package metadata, and release checks for the Compose workflow
Disclosure: TKEN is an independent third-party API gateway and is not officially affiliated with Open WebUI, LiteLLM, OpenAI, or other model providers.
v0.20.0 LlamaIndex OpenAI-compatible example
v0.20.0 adds a LlamaIndex integration pack for developers wiring TKEN into OpenAI-compatible LlamaIndex apps:
- guide for llama-index-llms-openai-like with TKEN api_base
- runnable Python example for a direct LLM smoke test before query engines or agents
- guardrails for embeddings, retrievers, function calling, structured output, low-limit agent tests, and redacted evidence
- README, docs navigation, tool integration notes, package scripts, and release checks for LlamaIndex adoption
Disclosure: TKEN is an independent third-party API gateway and is not officially affiliated with LlamaIndex, OpenAI, or other model providers.
v0.19.0 Vercel AI SDK and LangChain examples
v0.19.0 adds framework integration examples for developers wiring TKEN into Vercel AI SDK and LangChain apps:
- Vercel AI SDK guide using @ai-sdk/openai-compatible with the TKEN base URL
- LangChain guide for JavaScript @langchain/openai and Python langchain-openai
- runnable Node.js examples for AI SDK and LangChain JS
- runnable Python LangChain example
- rollout guardrails for server-side keys, /v1/models validation, non-streaming first checks, streaming/tool/structured-output follow-up tests, and redacted evidence
- README, docs navigation, tool integration notes, package scripts, and release checks for framework-level adoption
Disclosure: TKEN is an independent third-party API gateway and is not officially affiliated with Vercel, the AI SDK project, LangChain, OpenAI, or other model providers.
v0.18.0 SDK capability smoke tests
v0.18.0 adds OpenAI SDK capability smoke tests for developers validating optional features before routing production traffic through TKEN:
- Node.js and Python SDK scripts for streaming, JSON-mode responses, and tool-call checks
- guide for running one capability mode at a time after /v1/models and non-streaming chat pass
- decision table for when basic chat works but streaming, JSON parsing, or tool calls fail for a selected model route
- safe evidence template that avoids exposing keys, account IDs, private prompts, usage records, or provider dashboards
- README, docs navigation, compatibility notes, migration notes, and release checks for SDK feature rollout
Disclosure: TKEN is an independent third-party API gateway and is not officially affiliated with OpenAI or other model providers. Feature behavior can vary by selected model, provider route, account limits, and provider status.
v0.17.0 MCP host gateway config review
v0.17.0 adds an MCP host gateway config review pack for developers connecting MCP-capable hosts or agent apps to TKEN as an OpenAI-compatible model gateway:
- guide that separates MCP server config from model-provider base URL and API-key settings
- JSON review profile for host owner, model gateway env vars, selected model, MCP servers, permission plan, limits, logging, and rollout gates
- preflight sequence for /v1/models, one non-streaming chat request, one no-tool host prompt, then one read-only MCP tool
- failure map for gateway auth/base URL issues, host provider mismatch, MCP server/tool permission problems, runaway agent spend, and secret exposure
- README, docs navigation, and release checks for MCP-capable host configuration review
Disclosure: TKEN is an independent third-party API gateway and is not officially affiliated with OpenAI, Anthropic, MCP hosts, IDE vendors, model providers, or the Model Context Protocol project.
v0.16.0 - Endpoint preflight playbook
v0.16.0 adds an endpoint preflight playbook for developers validating TKEN before SDK, UI, proxy, agent, or CI rollout:
- recommended sequence for confirming the base URL includes /v1, running /models only, selecting a returned model ID, and running one non-streaming chat completion
- output interpretation for 401, 404, empty model lists, model-not-found errors, 429/quota, timeouts, 5xx responses, and response-shape mismatches
- safe evidence template for GitHub issues, team rollout notes, and deployment handoffs without exposing keys, account IDs, private prompts, usage records, or provider account screens
- CLI examples for
node tools/endpoint-tester.mjs --skip-chat --jsonand one selected-model chat check - README, docs navigation, tools documentation, and release checks for preflight adoption
Disclosure: TKEN is an independent third-party API gateway and is not officially affiliated with OpenAI or other model providers.