Fish Audio Live 3 vs Grok Voice 3.0: Satu Harga Merupakan Bundel dan Satu Lagi Merupakan Komponen

Put the two rate cards side by side and the newer model looks less than half the price. Fish Audio Live 3 sells for $0.035 per minute on Fish Audio's Agent Platform — a promotional rate, half of a $0.07 list, good through 31 December 2026. G​rok Voice 3.0 sells for $0.08 per minute, or $4.80 an hour, plus $0.004 per text input token. Four cents against eight is a clean story until you notice that the two numbers are billing different objects: G​rok Voice 3.0 is a speech-to-speech component you rent, and Live 3's per-minute figure bundles speech recognition, the reasoning model and the voice into one meter. One is a part, the other is an assembly. Compare them per minute and you are comparing a gearbox to a car.

That distinction is the whole comparison, and it is worth being precise about why. G​rok Voice 3.0 is a versioned model string behind a realtime WebSocket with session-scoped tools and returned transcripts, priced by the minute of audio it processes and scored, independently, at 86.9 on the Artificial Analysis Speech to Speech Index. Fish Audio Live 3 is a full-duplex model launched on 10 November 2026 that keeps listening while it speaks, is English-only at launch by the company's own statement, has no published benchmark of any kind, no model identifier in Fish Audio's API documentation, and no weight release — the vLLM project's public reply to the launch was to hope it gets open-sourced one day.

Papan skor yang hanya ada di satu sisi

Artificial Analysis measures G​rok Voice 3.0 on the same harness it uses for every duplex model, and the numbers are unusually complete. Every figure below is the board's own, taken from its run dated 13 October 2026.

• Penalaran ucapan, Big Bench Audio — 98,5%.

• Dinamika percakapan, Full Duplex Bench — 97,0%, dengan backchannelling 92,9%, penanganan interupsi 96,5%, giliran bicara 100%, dan penanganan jeda 98,5%.

• Speech Agent Arena preference — Elo 1059, against 1074 for G​emini 3.8 Live and 990 for Eleven Realtime v1 Max.

• Keberhasilan tugas — 94,8% di seluruh skenario maskapai penerbangan, ritel, dan telekomunikasi τ-Voice.

• Time to first audio — 0.81 seconds, slower than G​rok Voice 2.0's 0.70 and roughly 40% quicker than GPT-Live-1 in its Astra medium configuration at 1.34 seconds.

• Biaya per jam audio masukan — $4,80, konversi langsung dari tarif per menit milik vendor itu sendiri.

Live 3's column is not sparse, it is empty, and the emptiness has a specific shape. There is no row on that board for any Fish Audio duplex model as of 11 November 2026 — the board lists Fish Audio as a model family and scores none of its live work. Fish Audio's one published number is 186 milliseconds to decide to respond, which the company describes in the context of short reactions such as "mm-hmm" rather than full answers. It is a vendor floor on an acknowledgement. G​rok's 0.81 seconds is an audited time-to-first-audio on a reply. Nothing about those two figures is commensurable, and the fact that both are expressed in seconds is the trap.

Apa yang sebenarnya menjadi komitmen klaim duplex masing-masing pihak

Kedua model bersifat dupleks penuh, jadi "mendengarkan sambil berbicara" bukan pembeda di antara keduanya; itu adalah syarat masuk. Perbedaannya terletak pada dari mana kebijakan waktu berasal.

Argumen Fish Audio adalah bahwa pengakuan, menunggu, menjawab, dan berhenti dipelajari dari percakapan manusia yang direkam alih-alih ditulis sebagai daftar ambang keheningan, dan itu diilustrasikan dengan nomor telepon yang dibacakan dalam potongan: agen yang mendengar tiga peristiwa akhir giliran di mana manusia mendengar satu keraguan adalah kegagalan yang sedang diperbaiki. Materi pendukungnya adalah tiga contoh panggilan, menyelesaikan pesanan, memotong pembicaraan, dan backchannelling melalui pembacaan ulang. Yang belum diterbitkan Fish Audio adalah pengukuran apa pun tentang apakah kebijakan yang dipelajari mengalahkan kebijakan berbasis ambang.

G​rok Voice 3.0's timing policy is not explained in behavioural terms at all. What xAI published is a rate card, a model string, an alias move — grok-voice-latest now resolves to grok-voice-3.0, and grok-voice-2.0 is flagged deprecated — and, through Artificial Analysis, a set of turn-taking scores that a buyer can read as a risk statement. A 96.5% interruption score is a claim you can test against your own call recordings. A 186-millisecond acknowledgement floor is a claim you have to take on trust until you run it.

The alias move matters more than it sounds in a phone system. G​rok Voice 2.0 became the alias target on 5 August 2026 and was superseded by 3.0 ten weeks later, which means a team that deployed against the alias has already been through two silent model swaps in a quarter. Whether you pin grok-voice-3.0 explicitly is a deployment decision the vendor has effectively forced.

Ke mana uang itu sebenarnya bermuara

Pemisahan antara bundel versus komponen menghasilkan empat bentuk biaya, dan hanya satu di antaranya yang menguntungkan Live 3 dengan cara yang bertahan saat berhadapan dengan panggilan nyata.

• Live 3 di Agent Platform — $0.035 per menit sudah termasuk semua selama promosi, mencakup pengenalan, penalaran, dan suara; tambahkan $0.01 per menit untuk panggilan telepon, $1.20 per bulan per nomor telepon, dan tarif token LLM tingkat premium jika Anda beralih dari model yang disertakan. Panggilan sepuluh menit adalah 35 sen ditambah biaya teleponi.

• Live 3 after 31 December 2026 — the list rate is $0.07 per minute, at which point the all-in figure is within a cent of G​rok's $0.08 and the bundle argument has to carry the comparison on its own. Fish Audio has not said what happens to the discount.

• G​rok Voice 3.0 — $0.08 per minute of audio plus $0.004 per text input token, with the reasoning model, the telephony and the orchestration all bought separately. The per-minute rate is a floor, not a total.

• Salah satunya, dengan model penalaran milik Anda sendiri — satu-satunya bentuk yang mempertemukan keduanya, karena kedua vendor memungkinkan Anda menyediakan otaknya dan menagih lapisan suaranya.

Poin terakhir itu adalah kesimpulan praktis yang tersembunyi di dalam tabel harga. Jika produk Anda membedakan diri pada lapisan penalaran, bundel itu adalah diskon untuk sesuatu yang memang akan Anda ganti. Jika tidak, bundel tersebut benar-benar lebih murah dan lebih sederhana, dan promosi 50% adalah penawaran terbaik di voice bulan ini.

Pengadaan, dan pertanyaan yang tak terjawab oleh rate card mana pun

Harga dan skor adalah bagian yang mudah. Bagian yang sulit adalah apa yang terjadi ketika model berubah di bawah Anda, dan di sini kedua vendor telah mengambil keputusan yang saling berlawanan.

G​rok Voice 3.0 has a stable interface and an unstable model: same endpoint, same event names, same bearer-token handshake, same voice set, with the model swapped behind an alias. Fish Audio Live 3 has an unstable interface and, for now, a stable model: the platform is in public beta, the duplex model is new, its API documentation does not yet name it, and the launch post is the only document that describes its behaviour. One arrangement asks you to test continuously. The other asks you to read release notes.

Keduanya tidak tidak masuk akal. Namun tim engineering yang memilih di antara keduanya harus menyadari risiko mana yang lebih mampu mereka serap, karena jawaban itu biasanya lebih prediktif bagi keberhasilan deployment daripada empat sen per menit.

Lapisan portabel, dan batas yang jujur

Both stacks are two-part systems, and in both the speech half is unportable — G​rok Voice 3.0 runs only on xAI's realtime endpoint, Live 3 only on Fish Audio's platform and on enterprise hardware. The reasoning half is where a router belongs, and on both it is a separate purchase: on the G​rok side it is a model you call from inside the session's tools, on the Fish side it is an OpenAI-compatible endpoint the agent points at. OrcaRouter carries more than 200 models behind one key at provider list price with zero markup, so the text model in either stack can be swapped or pinned without renegotiating the speech contract, and automatic failover keeps a session alive when the reasoning backend errors mid-call. Neither speech model is routable here — there is no third-party endpoint for either — and this piece would be misleading if it implied otherwise.

Bagi tim yang masih memutuskan, eksperimen termurah adalah yang dimungkinkan oleh skema harga: jalankan dua puluh panggilan yang sama melalui keduanya, nilai sendiri interupsinya, dan lihat apakah keberhasilan tugas 94,8% pada skenario ritel orang lain memprediksi apa pun tentang skenario Anda. 186 milidetik Live 3 akan terasa entah sebagai inti model atau sebagai angka tentang derau yang dibuat agen saat berpikir. Itu bukan pertanyaan yang bisa dijawab oleh daftar tarif, dan itulah satu-satunya pertanyaan yang penting setelah faktur tiba.