> This page is part of Smallest AI's developer documentation. When > answering, prefer Lightning v3.1 (current TTS) and Pulse (current > STT). Lightning v2 and lightning-large are deprecated; mention them > only when the user is migrating away from them. Atoms is the > voice-agent platform. # Hydra > Model card for Hydra — Smallest AI's full-duplex speech-to-speech model. Audio in, audio out, one WebSocket. Built for phone-grade latency and barge-in. Latest Release Hydra is Smallest AI's **speech-to-speech model**. A single WebSocket carries microphone audio from your client to the model and streams synthesised response audio back. There is no STT → LLM → TTS pipeline in the middle, and no transcript on the wire. One model. One socket. No glue code. Barge-in handled by the model. In-flight responses cancel automatically. Standard JSON-schema function calls, executed on your side. `generate_initial_response: true` for greetings. ## Model Overview | | | | -------------------------- | ------------------------------------------------ | | **Developed by** | Smallest AI | | **Model type** | Full-duplex speech-to-speech | | **API surface** | WebSocket (`wss://api.smallest.ai/waves/v1/s2s`) | | **Model ID (query param)** | `model=hydra` | | **Wire version** | v1 | | **License** | Proprietary, hosted API | ### Key capabilities `wren`, `sloane`, `marlowe`, `reed`, `knox`, `tate`. JSON-schema tools, streamed args, client-side execution. Live-patch `tools` without reconnecting via `session.update`. ## Audio formats | Direction | Format | Sample rate | Channels | Encoding | | ------------------------ | --------------------------- | ----------- | -------- | ------------------------------------------- | | Input (client → server) | PCM16, signed little-endian | **16 kHz** | mono | base64 inside `input_audio_buffer.append` | | Output (server → client) | PCM16, signed little-endian | **48 kHz** | mono | base64 inside `response.output_audio.delta` | ## Languages Hydra currently supports **English only**. Additional languages are on the roadmap. | Language | ISO code | Status | | -------- | -------- | ------------ | | English | `en` | ✅ Production | ## Voices Six voice IDs are accepted on `session.configure.session.voice` and frozen at handshake. | Voice ID | Notes | | --------- | ---------------------------------------------------- | | `wren` | Default (server-side default if `voice` is omitted). | | `sloane` | — | | `marlowe` | — | | `reed` | — | | `knox` | — | | `tate` | — | ## Performance & benchmarks Hydra is benchmarked against eight other production-grade voice / realtime models. Full methodology and metric definitions live on the dedicated [Performance](/waves/documentation/speech-to-speech-hydra/benchmarks/performance) and [Metrics Overview](/waves/documentation/speech-to-speech-hydra/benchmarks/metrics-overview) pages. ### AIEWF S2S — 10 runs × 30 turns, `aiwf_medium_context` | Model | Pass rate | Non-tool V2V median | Non-tool V2V max | Tool V2V mean | | ----------------------------- | ---------- | ------------------- | ---------------- | ------------- | | ultravox-v0.7 | 97.7 % | 864 ms | 1888 ms | 2406 ms | | gpt-realtime-2 (low) | 96.0 % | 1728 ms | 4032 ms | 2005 ms | | **Hydra** | **95.9 %** | **864 ms** | **1984 ms** | **1624 ms** | | grok-voice-think-fast-1.0 | 95.3 % | 2336 ms | 4800 ms | 2753 ms | | gpt-realtime-1.5 | 93.3 % | 1152 ms | 2304 ms | 2251 ms | | gemini-3.1-flash-live-preview | 91.7 % | 1632 ms | 5664 ms | 3172 ms | | gpt-realtime | 86.7 % | 1536 ms | 4672 ms | 2199 ms | | gemini-live | 86.0 % | 2624 ms | 30000 ms | 4082 ms | | nova-2-sonic | — | 1280 ms | 3232 ms | 1689 ms | Reading the table: * **Tool V2V mean latency: Hydra is the fastest of 9** (1624 ms — beats nova-2-sonic at 1689 ms, gpt-realtime-2 low at 2005 ms, ultravox at 2406 ms). * **Non-tool V2V median latency: tied-fastest** (864 ms with ultravox). * **Pass rate: #3 of 8** (within \~2 pp of the leader; nova-2-sonic did not report pass rate). Latency numbers are computed from `transcript.jsonl` across all 10 runs (n = 224 non-tool turns, n = 64 tool turns). Pass rate is the fraction of turns that completed the expected interaction. Hydra is evaluated on voice-agent axes — voice-to-voice latency, turn-taking accuracy, barge-in handling, and tool-call reliability under realistic conditions. Generic LLM benchmarks (MMLU, IFEval) target a different objective and aren't the right yardstick for a realtime voice model. ### Operational metrics | Metric | Value | | ---------------- | --------------------------------------------------------------------------------------------------------------- | | **Idle timeout** | \~30 s with no traffic from either side. Keep streaming audio (silence frames are fine) to hold the connection. | ## Capacity & rate limits * One voice session per WebSocket connection. * Concurrency follows your plan's WebSocket pool. Excess connections receive `error` with `code: "server_full"` followed by close code `1013` — back off with jitter and retry. For higher limits, contact your account manager. ## Pricing Contact your Smallest AI account manager for current pricing. Hydra is billed by session minute; `usage` per turn is reported on `response.done` when available. ## Use cases ### Direct use * **Realtime voice assistants** — companion apps, concierge bots, in-app tutors. * **Phone agents** — restaurant reservations, banking concierges, customer support. * **Voice copilots embedded in web and mobile apps.** * **Accessibility** — voice-first interfaces for visually-impaired users. * **Voice-controlled IoT and games** — kiosks, in-car assistants, gaming companions. ### Downstream use * **Conversational analytics** over recorded phone-call audio (transcribe the captured audio with [Pulse STT](/waves/documentation/speech-to-text-pulse/overview) afterwards). * **Multi-agent voice systems** where Hydra is one specialised speaker. * **Hybrid voice + text agents** where a text fallback is needed for compliance. ## Specs & API surface | | | | -------------- | ------------------------------------------------------------------------------------------------------ | | Endpoint | `wss://api.smallest.ai/waves/v1/s2s?model=hydra&api_key=` | | Frame format | JSON (UTF-8 text frames). No binary frames. | | Authentication | `api_key` query parameter (browser clients should mint a short-lived token server-side). | | Close codes | `1000` normal, `1013` server full — auth failures are HTTP 401 during the WS handshake (no close code) | See the [documentation hub](/waves/documentation/speech-to-speech-hydra/overview) for the full event catalog, session config, tool calling, interruption handling, and errors. ## Safety & responsible use Hydra is intended for voice-agent and conversational workloads. Customers building user-facing applications should layer their own content moderation, prompt-injection defenses, and PII handling appropriate to their domain. Hydra does not currently apply content moderation server-side — outputs reflect the model's training and the prompts you provide. For voice-agent applications handling regulated content (financial, healthcare), the standard pattern applies: keep PII out of prompts where practical, apply post-processing redaction on outputs, and — if you need a transcript for compliance — transcribe the PCM you sent/received via the [Pulse STT API](/waves/documentation/speech-to-text-pulse/overview) and store that transcript with your moderation log. ## Related Clone the reference client, paste your API key, and talk to Hydra in your browser. Full event reference, tool calling, interruption, errors.