# Rime Prism (slim)

> Prism is Rime's speech-to-speech model for voice agents. It takes caller audio over one WebSocket, decides when the caller has finished, calls the developer's tools and replies in speech with matching text, in English.

- Full: https://www.anchorterminal.com/tools/rime-prism.md (~8,150 tokens) · this version ~1,680 tokens · JSON https://www.anchorterminal.com/tools/rime-prism.json · canonical https://www.anchorterminal.com/tools/rime-prism
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-10-29 · pricing https://www.anchorterminal.com/tools/rime-prism#pricing · reviews https://www.anchorterminal.com/tools/rime-prism#reviews
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

**C · 57.9/100 · rank #686 of 1222 · #14 in Conversational voice agents · not agent-ready · confidence medium**


Assessment: Prism's WebSocket API is well documented for a launch-day product, with typed events, 41 error codes that say what to do and a published per-minute price with free monthly usage. It has no Prism component or history on Rime's status page, no published concurrency limit, alpha SDKs, English only, and no data-handling statement that covers caller audio.

## Facts

- Kind: HTTP API · vendor: Rime Labs · category: Conversational voice agents · legal entity: Rime Labs, Inc. · provenance 81/100
- Endpoint: `wss://prism.api.rime.ai/v1/realtime` (websocket)
- Auth: API key · pricing: Freemium · x402: no · licence: Proprietary hosted service under Rime's terms. The Rime SDK used to call it (github.com/rimelabs/rime-sdk) is MIT.
- Probe metrics: not measured yet (probes haven't run)
- Architecture: One speech-to-speech model that hears the caller's audio, finds the end of each turn, writes its reply as text and speaks it. Half duplex at launch with interruption handling, per the pricing page. No choice of language model, no sampling settings and no text-only replies
- Endpoint: `wss://prism.api.rime.ai/v1/realtime`, in US East (us-east-2). One conversation per connection, OpenAI Realtime beta event names, Prism's own events and fields prefixed `prism`
- Audio: Headerless PCM16 mono, 16 kHz in and 24 kHz out, base64 in JSON frames. No mu-law or G.711, so phone calls need a bridge. The SDK accepts 8, 16, 24 or 48 kHz and converts it
- Languages and voices: English only. Four voices, `astra` (default), `aurora`, `brando` and `wawona`. Rime's text-to-speech voices don't work with Prism
- Tools: Functions in either OpenAI shape with JSON Schema parameters. `tool_choice` is `auto` only. The backend runs each call, returns a result and asks Prism to continue. Results time out after 90 seconds, and Prism can speak a filler phrase while a tool runs
- Limits: Context 65,536 tokens with oldest turns left out by default, replies up to 120 seconds or 512 tokens, 55 seconds per caller turn, 5 minutes idle, 1 MiB per client message. No concurrency limit per account was found
- Errors: 41 codes, each with a type, scope (event, utterance, response, tool round or session), the field at fault and a correlation ID, plus HTTP refusals (400, 401, 404, 503) and close codes (1000, 1001, 1011)
- SDKs: Rime SDK for Python (`rimelabs-sdk` 0.1.0a25), Node.js (`@rimelabs/sdk` 0.1.0-alpha.25), Go (v0.1.0-alpha.16) and Rust (0.1.0-alpha.14), all alpha. LiveKit Agents and Pipecat quickstarts, the Pipecat one on a pinned preview
- Deployment: Rime cloud on Free, Flex and Scale. The pricing page lists cloud, VPC or on-prem for Enterprise, while the docs describe only the cloud endpoint
- Assessment scope: Public pages only, read on 28 October 2026, the launch day. No account was created, no key was used and no call was made. Response time, accuracy and the benchmark results Rime cites were not tested
- Prices: Prism, Flex $0.08 per minute of call; Prism, Scale $0.064 per minute of call

- Scores: Reliability 41, Performance pending, Schema & documentation 69, Agent ergonomics 73, Security & auth 56, Payments & pricing 40, Task success pending, Maintenance & community 65, Transparency & trust 70 · total over the 7 assessed categories
- Why: Reliability, Graded as a hosted API on its launch day. · Schema & documentation, No OpenAPI or AsyncAPI file for the WebSocket API was found in llms.txt or the reference (0 of 25). · Agent ergonomics, Scored as a model API. · Security & auth, Graded as a hosted API. · Payments & pricing, No x402, MPP or L402 (0). · Maintenance & community, Read as a closed model API. · Transparency & trust, Closed service with clear terms from Rime Labs, Inc., which say Rime doesn't train on input or output except for a voice model the customer…
- Sources: 18, open questions: 7, both in the full twin
- Capabilities: voice.speech-to-speech, voice.agent, voice.tools
- JSON: https://www.anchorterminal.com/api/v1/tools/rime-prism.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/rime-prism.svg` or a link to https://www.anchorterminal.com/tools/rime-prism from a page on rime.ai or one of its subdomains, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.
- Support: support@rime.ai; https://docs.rime.ai/docs/prism/errors; more in the full twin

## Before you call it

1. Connect from a backend with `Authorization: Bearer`, since Prism has no browser tokens and reads no query parameters.
2. Send `session.update` before any audio, or every append fails with `session_not_initialized`.
3. Keep streaming audio, silence included, while Prism speaks, because it finds turn ends and interruptions from that audio.
4. Ask for the continuation after each tool result, since Prism doesn't resume on its own, and never rerun a call with an unknown outcome.
5. Pin the exact SDK version, because the SDKs are alpha and a new release can change the API.

## Connect

```bash
uv add "rimelabs-sdk==0.1.0a25"
```
Full config and headless snippets are in the full page. Through letme (picks today, calling later): https://letme.dev/rime-prism

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| LiveKit Agents | BB | 72.5 | voice.agent, voice.speech-to-speech, voice.tools | https://www.anchorterminal.com/tools/livekit-agents.min.md |
| Retell AI API + MCP | B | 69.2 | voice.agent, voice.speech-to-speech, voice.tools | https://www.anchorterminal.com/tools/retell-ai.min.md |
| Pipecat | B | 69.1 | voice.agent, voice.speech-to-speech, voice.tools | https://www.anchorterminal.com/tools/pipecat.min.md |
| GPT-Live | B | 66.9 | voice.speech-to-speech, voice.agent, voice.tools | https://www.anchorterminal.com/tools/gpt-live.min.md |
| Vapi API + MCP | B | 63.7 | voice.agent, voice.speech-to-speech, voice.tools | https://www.anchorterminal.com/tools/vapi.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)