GPT-Live-1 vs HeyGen Voice: Two Boards, Two Jobs, and One API That Already Pairs Them

Most write-ups of HeyGen Voice open with the leaderboard result, and the result is real: first place on Artificial Analysis's Controlled Voice arena at 1,198 Elo, ahead of Q​wen-Audio-3.1-TTS-Plus at 1,186 and Eleven v4 Turbo at 1,162. G​PT-Live-1 is not on that board and never will be, because the board measures how a voice sounds and G​PT-Live-1 is the thing that decides when to stop talking. The more useful fact is that HeyGen's own engineers already know this: in September 2026 they shipped an open-source real-time avatar stack that uses G​PT-Live-1 as the voice and conversation layer and their own avatars around it. The two products are not competitors. They are two halves of a system that one of the two vendors has already assembled in public.

What each one actually is

HeyGen Voice is HeyGen's in-house text-to-speech model, sitting inside a v3 API that also exposes video and avatar rendering. Two model IDs are published — heygen-voice-1, covering instant and professional clones, and heygen-voice-1-turbo, the same engine restricted to instant clones at lower latency. Internally the engine is called orca, which is the value you filter on in a voices query. It takes text and returns audio at 44.1 kHz mono WAV, either as a one-shot request (POST /v3/models/audio/tts) or as a server-sent-event stream with base64 WAV parts and optional character-alignment events (POST /v3/models/audio/tts/stream). It does not hear anything.

G​PT-Live-1 does nothing else. It handles a spoken conversation while a delegated backend model or agent reasons and calls tools, it listens while it speaks, and it runs on exactly one endpoint — v1/live/sessions — with Chat Completions, Responses, Realtime and Batch all unsupported. Audio and text in, audio and text out, images and video out of scope, knowledge cut-off 31 July 2025. Voice sessions bill at $0.05 per minute, by the second, with backend usage charged separately at normal model rates.

The architectural consequence is the whole comparison. On a HeyGen-style pipeline, the reasoner sits between the microphone and the voice. On G​PT-Live-1, the voice model is the pipeline and the reasoner hangs off it.

The leaderboard result, and the boundary around it

Artificial Analysis runs two speech boards and they answer different questions. The Provider Voice arena lets each vendor submit its own chosen voices — the polished demo voices a company picks to represent itself — and ranks them on how good those sound. The Controlled Voice arena substitutes a fixed set of eight cloned reference voices, four US and four UK, so every engine says the same words in the same voice under the same conditions. It is the closest thing the industry has to a like-for-like engine test, and it is the board that matters if you plan to clone a voice and hand it to a TTS model.

HeyGen Voice leads it. From the October 2026 capture:

• Controlled Voice Elo — HeyGen Voice 1,198 over 1,765 samples, against Q​wen-Audio-3.1-TTS-Plus at 1,186 over 2,097, Eleven v4 Turbo at 1,162, Eleven v4 at 1,159 and Cartesia Sonic 3.6 at 1,149.

• Provider Voice Elo — Eleven v4 Turbo leads at 1,330, Eleven v4 at 1,329, Q​wen-Audio-3.1-TTS-Plus at 1,296, Cartesia Sonic 3.6 at 1,285, Gemini 3.8 Flash TTS at 1,275. HeyGen Voice does not appear on this board.

• Price, on the arena's own basis — $30.00 per million characters, against Q​wen's $19.30 and Eleven v4 Turbo's $40.00. HeyGen publishes the same $30.00 per million characters on its own product page with the line "no packages, no minimums, no tiers."

That absence from the wider board is not a quality signal. Artificial Analysis notes entries can be withheld for insufficient votes, and a model weeks old with one differently-scored arena behind it has not accumulated the blind-listener volume the ninety-plus-model board requires. The correct reading is narrow and useful: HeyGen Voice is currently the top-scoring engine on the controlled, cloned-voice comparison, and an unknown quantity against the broader field. Anyone quoting "the best TTS model in the world" off that number is quoting a smaller board than the sentence implies.

What HeyGen gives you to build with

The engine is more flexible than a "TTS model" label suggests, and three details matter for anyone choosing between these two products.

• Engine selection per voice. A GET /v3/voices call takes an engine filter whose values include orca for HeyGen Voice, starfish, and a pass-through to ElevenLabs voices. HeyGen does not make you take its own model — it hedges, in the most literal sense, by letting you route to a competitor's engine through its own API. Anyone reading the 1,198 Elo as "HeyGen's voice is best" should note that it describes one of three paths the same endpoint offers.

• Cloning in two modes. Instant Clone takes one recording, uses the first three minutes, cannot be retrained, and is documented as free during preview. Professional Clone trains on one to ten recordings totalling twenty-plus minutes, returns 202 Accepted and polls until active, is retrainable, and occupies a purchased slot. A similarity control runs one to four, an expressiveness boost from 0.0 to 1.0 applies to instant voices only, and break tags up to five seconds are available on the professional tier.

• Voice design that is reproducible. POST /v3/voices generates up to three ranked candidates from a description capped at 1,000 characters, with an optional gender filter, an optional BCP-47 locale and a seed — and HeyGen documents that the same prompt plus the same seed always returns the same voices. Speech text caps at 5,000 characters and the endpoint runs at 30 requests a minute on pay-as-you-go, 120 on Enterprise.

• Language coverage of 39 base codes for both cloning and speech, with regional tags resolving to their base rather than switching accent. The platform-level "175+ languages and dialects" figure circulating on HeyGen's marketing pages is not this model's coverage.

One practical warning about that language list: several regional locales that the avatar engines accept do not select a regional accent in the voice engine. If your product localises, test the locale you need rather than assuming the tag carries.

The part of G​PT-Live-1 that HeyGen's own stack demonstrates

On 16 October 2026 O​penAI released prompt-to-voice to all API customers: a POST to /v1/audio/voices with type: "prompt", a name, the description and "model": "auto". Creating a voice is free; using it bills at normal session rates. The response is metadata only — an ID, a name, a creation timestamp, your prompt — with no audio and no preview, so auditioning costs a session. The steerable qualities are accent, age, presentation, pitch, timbre, texture, loudness, speech rate and distinctiveness, all expressed in ordinary language with no numeric parameters, and O​penAI documents that generated voices are not guaranteed to be unique.

Put that next to HeyGen's design endpoint and the difference is not quality, it is engineering discipline. HeyGen gives you up to three candidates, a seed that makes results reproducible and a preview audio URL per voice. O​penAI gives you one voice, no preview and no uniqueness guarantee, under a 100-voice organisation cap that also applies to voices generated on behalf of your customers. If a voice is a build artefact you will need to regenerate, only one of these two surfaces is designed for that.

The preset libraries diverge just as sharply. G​PT-Live-1's Live voice guide lists twenty-one named presets plus marin as default; the nine added on 16 October are Nuri, Juni and Flitz in German, Brise and Sillage in French, Nira in Hindi, Harema and Shida in Japanese, and Noeul in Korean — which leaves Hindi and Korean with one voice each, both feminine. HeyGen's catalogue is documented at 300-plus pre-built voices across dozens of languages. Those are not comparable numbers in kind, and for a product that needs a wide cast of distinct speakers, the difference decides the decision on its own.

Where the two systems meet: delegation, and what it costs

Both sides now offer a full conversational loop, and both put the reasoner somewhere you can control.

HeyGen's LiveAvatar ships in FULL mode, where HeyGen runs speech recognition, the language model, text-to-speech, turn-taking, memory and contexts, and LITE mode, where you keep your own agent and use HeyGen for the avatar and voice layer. G​PT-Live-1 does the same thing with a different boundary: Responses delegation, where O​penAI runs the backend model, or client delegation, where you point the session at your own harness, agent or service.

That shared shape is what makes the cost question identical on both sides, and it is the question most comparisons skip. On G​PT-Live-1, full active session time is billable including while the backend works. O​penAI's own cost guidance makes the arithmetic blunt: using a faster backend with parallel tool calls so the user finishes a minute earlier saves $0.05 in voice charges. The voice rate is a constant; session length is the dial, and on HeyGen's FULL mode the same economics apply to whatever reasoner sits in the loop. A TTS line item does not appear in a conversational product's budget in the way a procurement sheet implies — the reasoning minutes do.

That is the one place a routing layer changes the shape of the build rather than the price of a line item. Client delegation turns the backend into a variable you can route per conversation instead of a fixed dependency. OrcaRouter fronts 200-plus models at provider list price with no markup, which matters most exactly when a vendor cuts a price — the cut is live here the same day rather than at renewal. Automatic failover keeps a provider outage out of a live call, and the routing DSL lets you compose the reasoning step from several models rather than staking the session on one. To state the boundary plainly: OrcaRouter hosts neither G​PT-Live-1 nor HeyGen Voice, and neither appears in our catalogue. What it routes is the reasoning behind a delegated session.

How to decide this week

Write down which board your product is judged on, because these two models do not overlap there. If a human listens to the output asynchronously — a marketing video, a training module, an audiobook, a cloned brand voice reading many lines — you are on the Controlled Voice board, HeyGen Voice currently leads it, and its $30.00 per million characters sits mid-field against Q​wen's $19.30 and Eleven v4 Turbo's $40.00. Test it against both of those in the same arena, because two of the three are cheaper.

If a human talks back in real time, the Controlled Voice board has nothing to say to you. The question becomes who owns the turn-taking loop, what the backend costs per session, and whether the voice model is the pipeline or a component inside one. HeyGen's answer is LiveAvatar with an agent you bring; O​penAI's answer is a session that delegates. Both are viable, and the fact that HeyGen's own reference stack uses G​PT-Live-1 for the conversational half is the clearest evidence available that these are not substitutes.

What not to do is quote "first place on Artificial Analysis" as though it settled a conversation product's decision. It settled a rendering question on eight cloned voices. The other question has not been asked on that board at all.

None of that is hypothetical: the documented AI failures and agent incidents behind releases like this one are collected in the OrcaRouter AI Incident Archive, where each entry carries its own sources, severity, confidence grade and disputed flag.