Chirp 2 vs Chirp 3
Which one, when — curated verdict, not a benchmark table
Same $0.016 per audio minute, so the newer chirp-3 is the default: it adds speaker diarization in Recognize and BatchRecognize across 14 languages, and publishes 29 GA plus 82 preview locales where chirp-2 does not diarize and limits StreamingRecognize to 16 locales. The trade is timestamps and length — chirp-2 offers word-level timestamps and batch jobs from 1 minute to 8 hours, while chirp-3 lists word timestamps as unsupported and caps batch at 1 hour. Pick chirp-2 when you need word timings or very long files.
Pricing
Rates from the live catalog at build time; each model page carries the current card.
Where they sit — price per audio minute across all 11 speech-to-text models on this billing unit (log scale)
Capabilities
Specs
| Chirp 2 | Chirp 3 | |
|---|---|---|
| Input modalities | audio | audio |
| Output modalities | text | text |
| Released | — | 2025-10-13 |
| Limits | Auto-detected audio decoding sync Recognize <1 min, BatchRecognize 1 min–8 hrs, StreamingRecognize for real-time optional word-level timestamps language-agnostic transcription (the model infers the prevalent spoken language and transcribes in it) no standalone language detection no speaker diarization speech translation supported | Auto-detected audio decoding sync Recognize <1 min, BatchRecognize 1 min–1 hr (<=20 min with word timestamps), StreamingRecognize for real-time speaker diarization in BatchRecognize and Recognize (14 languages) utterance-level timestamps (StreamingRecognize only), word-level timestamps listed as unsupported language-agnostic transcription 29 GA + 82 preview locales |
| Languages | Varies by method: BatchRecognize offers the most extensive coverage and Recognize is "on par with Chirp" StreamingRecognize is limited to 16 locales (Chinese Simplified/Traditional, Cantonese, English AU/IN/GB/US, French CA/FR, German, Italian, Japanese, Korean, Portuguese (Brazil), Spanish ES/US) | 29 GA + 82 Preview locales (111 total) across StreamingRecognize, Recognize and BatchRecognize diarization covers 14 of them |
Specs are transcribed from each vendor’s documentation; a row a vendor does not publish is left out rather than inferred. Full sources: Chirp 2 · Chirp 3
Switch between them with one line
Both ids are in every tab below — the highlighted pair of lines is the only edit. Same endpoint, same key, same request shape.
from openai import OpenAI
client = OpenAI(
base_url="https://synthorai.io/v1",
api_key="sk-syn-...",
)
resp = client.audio.transcriptions.create(
model="chirp-2",
# model="chirp-3", # uncomment this line, comment the one above
file=open("meeting.mp3", "rb"),
language="en",
)
print(resp.text)import OpenAI from "openai";
import fs from "node:fs";
const client = new OpenAI({
baseURL: "https://synthorai.io/v1",
apiKey: "sk-syn-...",
});
const resp = await client.audio.transcriptions.create({
model: "chirp-2",
// model: "chirp-3", // uncomment this line, comment the one above
file: fs.createReadStream("meeting.mp3"),
});
console.log(resp.text);curl https://synthorai.io/v1/audio/transcriptions \
-H "Authorization: Bearer sk-syn-..." \
-F model="chirp-2" \
# -F model="chirp-3" \ # uncomment this line, comment the one above
-F file=@meeting.mp3package main
import (
"context"
"fmt"
"os"
"github.com/openai/openai-go/v3"
"github.com/openai/openai-go/v3/option"
)
func main() {
client := openai.NewClient(
option.WithBaseURL("https://synthorai.io/v1"),
option.WithAPIKey("sk-syn-..."),
)
f, _ := os.Open("meeting.mp3")
resp, _ := client.Audio.Transcriptions.New(context.TODO(), openai.AudioTranscriptionNewParams{
Model: "chirp-2",
// Model: "chirp-3", // uncomment this line, comment the one above
File: f,
})
fmt.Println(resp.Text)
}import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.audio.transcriptions.*;
import java.nio.file.Paths;
OpenAIClient client = OpenAIOkHttpClient.builder()
.baseUrl("https://synthorai.io/v1")
.apiKey("sk-syn-...")
.build();
Transcription resp = client.audio().transcriptions().create(
TranscriptionCreateParams.builder()
.model("chirp-2")
// .model("chirp-3") // uncomment this line, comment the one above
.file(Paths.get("meeting.mp3"))
.build()).asTranscription();
System.out.println(resp.text());FAQ
Which is cheaper, Chirp 2 or Chirp 3?
They list the same per audio minute ($0.016), so price does not decide this one — see the specs and capabilities below.
Can I A/B test Chirp 2 against Chirp 3 without two integrations?
Yes. Both are served through the same OpenAI-compatible endpoint with one API key — switching is a one-line model-string change, so you can route a fraction of traffic to each and compare bills directly.
Do Chirp 2 and Chirp 3 support speaker diarization?
The capability table above answers this per model, straight from each vendor’s documentation — diarization, streaming and timestamps are listed separately because models differ on all three.