v1.27.0Tera Router v1.27.0 is out: open source, GPL-3.0 licensed, self-hosted. Release notes

Every AI provider, behind one endpoint.

Tera Router is an open source AI gateway and LLM router. Speak OpenAI, Anthropic, or Responses — route across accounts with fallback, keys, guardrails, and usage analytics built in.

v1.27.0 · GPL-3.0 licensed · self-hosted · no per-token markup.

Routes to the accounts you already pay for — API keys, coding plans, or local models

  • OpenAI
  • Anthropic
  • Gemini
  • OpenRouter
  • NVIDIA NIM
  • Mistral AI
  • Together AI
  • Fireworks AI
  • Nebius Token Factory
  • Groq
  • Cline
  • Cloudflare AI
  • Meta Muse
  • OpenAI Codex
  • Exa
  • Tavily
  • Firecrawl
  • Brave Search
  • Serper
  • Voyage AI
  • Jina AI
  • ElevenLabs
  • Deepgram

Turn every account you own into one endpoint.

Connect accounts once, then let the planner handle the rest: priority order, cooldowns, fallback across providers, and full metering of what each key consumes.

providers in the catalog
20+
providers in the catalog
API dialects, one gateway
3
API dialects, one gateway

Your tools

native SDKs

  • Claude CodeAnthropic SDK
  • CursorOpenAI-compatible
  • Codex CLIResponses API
  • LangChainOpenAI SDK
  • curl & scriptsany HTTP client

Tera Router

plan · transform · meter

Your accounts

priority order

  • Anthropicpriority 1
  • Geminipriority 2
  • OpenAI Codexoauth session
  • OpenRouterfallback
  • Self-hosted vLLMcustom endpoint
  1. 01

    Connect providers

    Pick from the built-in catalog — API keys, OAuth sessions, or custom OpenAI- and Anthropic-compatible endpoints. Multiple accounts per provider, each with its own base URL and priority.

    openai · anthropic · gemini · openrouter …

  2. 02

    Route with fallback

    Every request enters one endpoint. The planner resolves accounts in priority order, cools down failures, and falls back across accounts and providers automatically.

    glm-4.7 → kimi-k2 → claude-sonnet (fallback)

  3. 03

    Observe everything

    Token usage is metered per request — including SSE streams — and lands in the admin dashboard: per key, per account, per model.

    GET /admin/usage → per-key analytics

Built for running, not just proxying.

Routing is the headline. The gateway also carries the operational load — access control, policy, and telemetry — in a single Go service you deploy once.

Streaming-first request log

SSE passes straight through for chat and messages, with token usage extracted from the stream itself — no buffering, no lost metrics.

Filter requests, keys, models…live
  • 17:02:14INFsearchexa · 200 · 8 results · 412 ms
  • 17:02:13INFresponsesgpt-5-codex → codex · stream · 22.7k tok
  • 17:02:12ERRguardrailkey tr_live_…9f2 · model not in allowlist
  • 17:02:11INFmessagesclaude-sonnet-4.5 → anthropic#2 · fallback · 200
  • 17:02:10WRNmessagesanthropic#1 429 · cooling down 60s
  • 17:02:09INFchatgemini-2.5-pro → gemini#2 · 200 · 1,204 tok

Three dialects, one gateway

The transform engine rewrites between OpenAI Chat, Anthropic Messages, and Responses, so a request in one dialect can land on a provider that speaks another.

  • OpenAI Chat Completions

    POST /v1/chat/completions

  • Anthropic Messages

    POST /v1/messages

  • OpenAI Responses

    POST /v1/responses

Fallback across accounts and providers

Several accounts per provider, tried in priority order. Failures cool down automatically and traffic moves to the next account in the chain.

  • anthropic#197.8%
  • gemini#2100%
  • openrouter99.6%
  • groq#1100%

Policy before traffic leaves your box

Scope every gateway key to specific models, cap its concurrency, and run requests through the guardrails engine — enforced centrally.

  • Per-key model allowlists
  • Guardrails policy engine
  • Concurrency limits per key
  • Quotas & budgets per account
  • Proxy pools for upstream egress

Bring every account you already have.

A built-in catalog of presets — base URL, dialect, and auth mode included — plus custom endpoints for anything OpenAI- or Anthropic-compatible.

Chat & LLM

13

Bring your own key — catalog presets ship the base URL and dialect.

  • OpenAI
  • Anthropic
  • Gemini
  • OpenRouter
  • NVIDIA NIM
  • Mistral AI
  • Together AI
  • Fireworks AI
  • Nebius Token Factory
  • Groq
  • Cline
  • Cloudflare AI
  • Meta Muse

Subscription / OAuth

1

Sign in with an existing ChatGPT subscription and route it over the Responses dialect.

  • OpenAI Codex

Search & embeddings

7

Served on /v1/search, /v1/web/fetch, and /v1/embeddings behind the same gateway keys.

  • Exa
  • Tavily
  • Firecrawl
  • Brave Search
  • Serper
  • Voyage AI
  • Jina AI

Voice

2

Text-to-speech and transcription on /v1/audio/speech and /v1/audio/transcriptions.

  • ElevenLabs
  • Deepgram

Custom

2

Any OpenAI- or Anthropic-compatible endpoint, with per-account base URLs.

  • OpenAI-compatible
  • Anthropic-compatible

One container, one base URL.

One container next to your stack, one base URL in every tool. From clone to first routed request in minutes.

01Run the gateway

bash
$ git clone https://github.com/montara-project/tera-router
$ cd tera-router/deploy
$ echo "APP_MACHINE_ID=1" > .env
$ echo "APP_SECRET=$(openssl rand -base64 48)" >> .env
$ docker compose up -d
✔ Container deploy-app-1  Started
✔ Gateway and dashboard on http://localhost:8080

02Point your SDK

python
client = OpenAI(
  base_url="http://localhost:8080/v1",
  api_key="tr_live_…",
)

Self-host it

deploy/docker-compose.yaml runs the gateway and dashboard as one image on port 8080.

GPL-3.0 licensed

The whole monorepo is GPL-3.0 — read it, change it, run it in production.

Bring your own stack

No account, no seat licence, no per-token markup — the gateway just routes.

Stop juggling SDKs, keys, and quotas.

Give every tool the same base URL and one key. Tera Router picks the account, handles the dialect, and shows you exactly what was spent.

  • GPL-3.0 licensed, free to self-host
  • Keys never leave your box
  • One Go binary, one container