# Speko

> Speko operates the hosted Router for managed STT, LLM, and TTS routing and publishes the open customer-side Gateway for provider-direct voice workloads.

- Console: https://platform.speko.ai
- Hosted Router: https://router.speko.dev
- Router OpenAPI: https://speko.ai/openapi.json
- RFC 9727 API catalog: https://speko.ai/.well-known/api-catalog
- Agentic Resource Discovery catalog: https://speko.ai/.well-known/ai-catalog.json
- Authentication and OAuth discovery: https://speko.ai/auth.md
- Router versioning and deprecation policy: https://speko.ai/developers/versioning
- Open customer-side Gateway: https://github.com/SpekoAI/gateway
- Hosted Agents API: https://api.speko.dev
- Public docs MCP (no authentication): https://speko.ai/.well-known/mcp
- Operational MCP (OAuth or bearer key): https://mcp.speko.ai/mcp
- Official CLI: https://speko.ai/developers/cli
- Benchmarks: https://benchmarks.speko.ai/
- Whole public-site corpus: https://speko.ai/llms-full.txt
- Router and Gateway scoped index: https://speko.ai/docs/llms.txt
- Developer-resource scoped index: https://speko.ai/developers/llms.txt
- Changelog feed (JSON Feed 1.1): https://speko.ai/changelog.json
- Changelog feed (RSS 2.0): https://speko.ai/changelog.xml

## When to use Speko

Use the hosted Router when an application wants Speko to operate a managed,
provider-neutral STT, LLM, or TTS data plane with typed requests, typed
responses, explicit route metadata, idempotent writes, and normalized errors.
Use the open Gateway when a LiveKit or Pipecat workload needs provider-direct
streaming, local BYOK credential custody, or a mixture of local and
Speko-managed routes. Use the hosted Agents API when the job is to create and
operate agents, sessions, phone numbers, or calls. These are distinct product
boundaries.

Use the public docs MCP at https://speko.ai/.well-known/mcp when an agent needs Speko
documentation without account access. Use https://mcp.speko.ai/mcp only for
authenticated account data or actions. Do not use Router as a hostname
replacement for an OpenAI SDK: Router has its own public OpenAPI and AsyncAPI
contracts.

## Hosted Router

Base URL: https://router.speko.dev

Authenticated operations send:

```text
Authorization: Bearer $SPEKO_API_KEY
```

Every POST and WebSocket upgrade also sends a caller-generated
`Idempotency-Key`. Reuse a key only for a byte-identical retry.

HTTP operations:

```text
GET  /openapi.json
GET  /v1/models
POST /v1/stt/transcriptions
POST /v1/tts/speech
POST /v1/llm/responses
```

Streaming speech uses WebSocket operations at `/v1/stt/stream` and
`/v1/tts/stream`; the AsyncAPI document and transport types live in the public
Gateway repository's `relayapi` package.

Routing is an explicit tagged object. Use
`{"mode":"auto","objective":"balanced"}` to let Router choose, or
`{"mode":"explicit","provider":"...","model":"..."}` to pin. The
objective is one of `balanced`, `quality`, `latency`, or `cost`.

Success responses identify the concrete route through structured response
bodies and the `Speko-Provider`, `Speko-Model`, `Speko-Region`,
`Speko-Request-ID`, and `Speko-Attempt-ID` headers where applicable. Every
non-2xx response uses one JSON envelope:

```json
{"error":{"code":"authentication_failed","message":"...","hint":"...","retryable":false,"request_id":"req_..."}}
```

The closed error vocabulary includes `invalid_request`,
`authentication_failed`, `route_not_found`, `method_not_allowed`,
`idempotency_conflict`, `rate_limited`, `concurrency_exhausted`,
`provider_error`, `provider_unavailable`, and `request_timeout`.

Every public Router response advertises the default edge policy as
`RateLimit-Policy: "relay-ip";q=2000;w=300`: 2,000 requests per five-minute
window for each public source IP. Organization and provider concurrency can
impose lower effective limits. Application-generated 429 responses also carry
`Retry-After`; wait at least that long and continue with exponential backoff.

The `/v1` path is the stable major version. Compatible additions may ship
within v1. Breaking changes require a new major path. Deprecations use RFC 9745
`Deprecation`, a `Link` with `rel="deprecation"`, a dated `Sunset` header,
migration instructions, and at least 90 days of notice.

## Open customer-side Gateway

Speko Gateway is not the hosted Router. It is the public customer-side runtime
at https://github.com/SpekoAI/gateway. It provides one local streaming protocol
across voice providers, keeps BYOK credentials in the Gateway process, and
ships native LiveKit and Pipecat integrations. Provider-direct voice traffic
can stay local while selected work uses Speko-managed routes. Read
https://github.com/SpekoAI/gateway/blob/main/docs/TRUST.md before choosing the
credential and process boundary.

## Pages

- [The Router for Voice AI](https://speko.ai/index.html.md): Hosted Router for managed STT, LLM and TTS routing, plus an open customer-side Gateway for provider-direct voice workloads.
- [Voice agent infrastructure for STT, LLM and TTS](https://speko.ai/voice-agent-infrastructure/index.html.md): Voice agent infrastructure across the hosted Speko Router, the open customer-side Gateway, and the separate hosted Agents API.
- [Voice AI models and benchmarks](https://speko.ai/models/index.html.md): Measured voice AI benchmarks across speech-to-text, LLM, text-to-speech, and speech-to-speech, with current Router route availability kept separate from historical evidence.
- [Speko Router API quickstart](https://speko.ai/docs/index.html.md): Call Speko Router's native model discovery, speech-to-text, LLM, and text-to-speech operations with typed responses.
- [Pricing](https://speko.ai/pricing/index.html.md): $0.09 per minute all-in on Speko infrastructure across speech-to-text, LLM and text-to-speech, or 5% on top of a provider's published rate when routed. $10 in credits on signup.
- [About Speko](https://speko.ai/about/index.html.md): Speko Group, Inc. builds the hosted Speko Router and the open customer-side Speko Gateway for real-time voice AI workloads.
- [Changelog](https://speko.ai/changelog/index.html.md): Dated Speko API, model, SDK, console, and documentation changes, with migration actions and effective dates for breaking, deprecated, and retired updates.
- [Speko developer resources](https://speko.ai/developers/index.html.md): Speko Router, Gateway, Agents API, OpenAPI, framework guides, MCP, CLI, and machine-readable discovery resources in one place.
- [Speko CLI](https://speko.ai/developers/cli/index.html.md): Official Speko command-line tools on npm for disclosed agent phone calls and local access to the hosted Speko MCP server.
- [Speko API versioning and deprecation policy](https://speko.ai/developers/versioning/index.html.md): Speko REST API versioning, backward-compatibility, deprecation, sunset, migration, and rate-limit conventions for durable agent integrations.
- [Speko Gateway for LiveKit](https://speko.ai/docs/livekit/index.html.md): Run the open Speko Gateway beside a LiveKit AgentSession for provider-direct voice legs and optional managed Router LLMs.
- [Speko Gateway for Pipecat](https://speko.ai/docs/pipecat/index.html.md): Add Speko Gateway's native STT and TTS services to a Pipecat cascade while keeping the rest of the pipeline unchanged.
- [Privacy Policy](https://speko.ai/privacy-policy/index.html.md): How Speko collects, uses, shares, and protects information when you use the Speko website, platform, APIs, SDKs, demos, and integrations.
- [Speko vs Bland AI](https://speko.ai/speko-vs-bland/index.html.md): A sourced comparison of Speko Router and Gateway with Bland AI across product scope, telephony, routing, runtime ownership, observability and pricing.
- [Speko vs Retell AI](https://speko.ai/speko-vs-retell/index.html.md): A sourced comparison of Speko Router and Gateway with Retell AI across product scope, telephony, routing, runtime ownership, observability and pricing.
- [Speko vs Vapi](https://speko.ai/speko-vs-vapi/index.html.md): A sourced comparison of Speko Router and Gateway with Vapi across product scope, telephony, routing, runtime ownership, observability and pricing.
- [Support](https://speko.ai/support/index.html.md): How to reach the Speko team, what to include so we can help on the first reply, and where to check whether something is already known to be down.
- [Terms of Service](https://speko.ai/terms-of-service/index.html.md): The terms that govern access to and use of Speko's voice routing platform, APIs, SDKs, and demos: accounts, acceptable use, billing, and liability.
- [Vapi Alternatives](https://speko.ai/vapi-alternatives/index.html.md): Sourced Vapi alternatives for voice agent teams: Speko Router and Gateway, Retell AI, Bland AI, Pipecat and LiveKit Agents, compared by product category.
- [Voice agent latency measurement framework](https://speko.ai/voice-agent-latency/index.html.md): A production framework for tracing voice agent latency across turn detection, streaming STT, LLM first token, TTS first audio and transport.
- [Voice agent platforms, runtimes and model routers](https://speko.ai/voice-agent-platforms/index.html.md): A sourced comparison of Vapi, Retell AI, Bland AI, Pipecat and Speko across telephony, provider choice, runtime ownership, operations, pricing and tradeoffs.
- [Find a voice](https://speko.ai/voices/index.html.md): Browse measured text-to-speech models and their naturalness evidence: 17 voice models on the English board and 9 more languages measured.