24 AUG 2026
Tingvar for developers
There is nothing to call yet. What exists is a connection contract worth arguing with before it is fixed, and one measured latency figure published with the method that produced it.
What exists today
| What | What it is |
|---|---|
| A published latency figure | Voice to voice, measured against a real media server, with the method, the sample size and the five stages it breaks into. It is on the evidence page, and it is published with the method rather than as a headline, because a latency number without one is a marketing adjective with digits in it. |
| A written connection contract | The transport, the auth model, the tool policy and the list of things the server may not do, below. It is worth reading before the server exists, because it is what you would be wiring into. |
| Nothing to call | There is no endpoint, no key issuance and no reference. Nothing on this page returns a 200 today, and the rest of it is written in the tense that fact deserves. |
Voice to voice
1471 ms
Median, on the bench in us-central1, with a real media server
MEASURED 2026-08-19, 122 readings
Real human sessions
0 sessions
Have taken place. Every measured turn so far used pre-rendered audio
MEASURED 2026-08-23, 1 reading
The left figure is voice to voice: you stop speaking, the first persona's audio starts. It was measured against a real media server rather than a simulation, and the method, the sample size and the five stages it breaks into are on the evidence page. The right figure is why the left one is not sufficient on its own.
What does not exist yet
| What | State and date |
|---|---|
| The MCP server | In development. No date yet. The connection contract below is published now so it can be argued with before it is fixed. Keys are issued from account settings and scoped to one workspace, and no account can create one today. |
| A REST API | Not started. No date yet. There is no endpoint, no request shape, no response shape and no reference at a stable address. Whatever is published is published in the same commit that makes it answer, and this page links to it on that day rather than before. |
| SDKs | Not started. No date yet. No client library is written for any language, and none is packaged or published under any name. A package on any registry that claims to be a Tingvar SDK today is not ours. |
| A public tool list with signatures | In development. No date yet. The names on the MCP page are a proposal and are not ratified. The list the server advertises is published in full, with each signature, in the commit that makes the server reachable, and it is versioned from that day. |
| Webhooks and unattended runs | Not started. No date yet. Submitting a proposal from an agent and receiving a recorded debate later is written down rather than offered, for the reason below the table. |
Webhooks and an async mode are on the roadmap, and they are last for a reason. An unattended room has the least evidence behind it, because interruption is what makes a room worth holding and an unattended room has none. It sits on the roadmap, with no date beside it there either.
The connection contract
The same rows the MCP page carries, from the same module, so the two cannot drift apart. Three of the six have no value yet and say so in the row. This page carries no server address, because none has been issued and a hostname printed before it resolves is the first thing a developer would try.
| Field | Value |
|---|---|
| Transport | Streamable HTTP. It is the transport a hosted server uses and the one every client on the list below speaks. |
| Server address | Not issued yet. This page carries no hostname until one resolves, and printing one early would be the first thing on this page a reader would try. |
| Authentication | An account bearer token, issued from the signed in account settings and scoped to one workspace. No key issuance exists yet, so there is nothing to create today. |
| Tools | The tool list is not frozen. It is published here in full, with each tool signature, in the same commit that makes the server reachable, and it is versioned from that day. |
| Rate limits | A room is a 45 minute live audio session with a real cost behind it, so the limit is a concurrency limit on rooms rather than a request rate. The number is set when the per session cost is known. |
| What the server may not do | It may not read your repository, your files or your environment. It receives the brief the agent sends it, and nothing else reaches us. |
How a client is configured, tool by tool and step by step
What a call consumes
A tool call consumes nothing on its own. It returns a join URL, and one session is consumed at the moment a person opens that URL and the room starts, against the plan on the account the token belongs to. Nothing an agent does on its own can spend a session.
Tingvar plans stop at their included allowance rather than charging past it. Both plans are capped rather than metered, so the monthly figure on the pricing page is the largest amount you can be charged in a month. There is no overage and no silent upgrade. A plan that quietly bills past its own ceiling is a trick rather than a business model. Your transcripts stay available and exportable.
Client compatibility
No client row is published here, because no client has been tested. A row is earned rather than asserted, and this is what earning one means:
- the client installed, at a version we write down;
- the operating system it ran on, written down beside it;
- that client connected to a live endpoint by a named tester;
- the result and the ISO date of the test recorded against the row.
No row can be added until an endpoint exists to connect to. There is nothing Tingvar specific in the connection, so a client that can already reach a hosted MCP server is likely to reach this one, and likely is not a row.
Two questions worth the space
Can I use it from a coding agent or an automation?
Tingvar exposes an MCP server, so an agent you already run can convene a room from where your proposal already lives. Claude Code, Cursor, Windsurf and Zed all speak MCP. The connection details, the authentication and an exact list of what exists today against what is still roadmap are on the MCP page, written in the tense each one deserves. A room convened this way is the same room you would join from a browser.
Could we build this ourselves with CrewAI or LangGraph?
You could build multi-agent text debate in an afternoon, and the hard part of Tingvar is none of that. The hard part is deciding which of three simultaneously willing speakers gets the floor, fast enough that the room still sounds like a conversation, without any of them talking over the human. Agent level turn-taking, where each agent decides for itself whether to speak, stops working at three agents. Our arbiter is a room level policy, and it is where the latency budget is spent.
IDLE
Nobody has used this yet
Zero real human sessions have taken place. Every measured turn to date used pre-rendered audio clips rather than a live human. What is built and verified is the decision layer: the turn-taking policy, the bid scoring and the configuration guards, with 619 automated tests passing and 81.78 % combined coverage. Everything there says the system works. Nothing there says anyone wants it, and being first is a real risk you are taking.
For a developer deciding whether to wire this into anything, that is the load bearing paragraph on the page.