Turn text into speech without a voice-platform contract

Send the text, a model and a voice; the audio comes back in the response. A voices endpoint on the same path lists what each model can speak with. The card prices a thousand characters of input on each model, which is the unit the payment layer meters — a longer text costs proportionally more, quoted before it is synthesised.

The flash-tier models are built for real-time voice agents and are the cheapest on the card; the higher tiers trade latency for expressiveness. Sound effects have their own endpoint and their own page.

In Claude Code, Claude Desktop, Cursor, Windsurf, Cline and any other MCP host, this is the blockrun_speech tool of the BlockRun MCP server — the agent calls it like any other tool and the result comes back in the conversation. In OpenClaw, Codex and Hermes, ClawRouter exposes the same catalog. In your own code, the Python and TypeScript SDKs wrap the endpoint above with types; any HTTP client works too. Every harness, with its setup →

No account and no API key stand between an agent and the first call: the request is quoted in dollars and the agent's client pays it. A team that prefers keys gets one, with a monthly invoice.

The direct vendors — ElevenLabs, OpenAI's TTS endpoint, Google Cloud Text-to-Speech, Azure Speech — each sell their own voices behind their own account. ElevenLabs in particular is priced as a subscription with a monthly character allowance, which is the wrong shape for an agent that speaks a few hundred characters at unpredictable times and nothing for days.

This endpoint carries more than one vendor's voice models, priced per thousand characters of input, with no monthly allowance to buy ahead. A voices endpoint on the same path lists what each model can speak with, so a caller picks a model and a voice in the same request without a second integration.

The trade-off is honest: the direct vendor's console has features — voice cloning, dubbing, projects — that an API for speech synthesis does not need and this page does not offer. What is here is the synthesis, on every voice model at once, billed for the characters you actually send.