Chirp 3 vs GPT-4o Transcribe Diarize
Which one, when — curated verdict, not a benchmark table
These two both do transcription with speaker diarization, but they bill in different units — chirp-3 at $0.016 per audio minute versus gpt-4o-transcribe-diarize at $6.25 per million audio input tokens plus $2.5 per million text input and output tokens — so no single conversion between them is honest. Pick chirp-3 for long or live jobs: BatchRecognize spans 1 minute to 1 hour, StreamingRecognize handles real-time, and it covers 29 GA plus 82 preview locales, though word-level timestamps are unsupported. Pick gpt-4o-transcribe-diarize for shorter files under 25MB where you want diarized_json with speaker labels and segment timestamps, within its 16000-token context and 2000-token output cap.
Pricing
| Chirp 3 | GPT-4o Transcribe Diarize | Δ | |
|---|---|---|---|
| Per audio minute | $0.016 | — | — |
| Audio input / 1M tokens | — | $6.25 | — |
| Text output / 1M tokens | — | $2.5 | — |
These two models bill in different units, so no Δ is shown — converting between them would require an assumption we have not measured. Each card is listed in its own unit above.
Capabilities
| Chirp 3 | GPT-4o Transcribe Diarize | |
|---|---|---|
| Speaker diarization | yes | yes |
| Streaming | yes | yes |
| Timestamps | yes | yes |
Specs
| Chirp 3 | GPT-4o Transcribe Diarize | |
|---|---|---|
| Input modalities | audio | text audio |
| Output modalities | text | text |
| Released | 2025-10-13 | 2025-10 |
| Knowledge cutoff | — | 2024-06 |
| Limits | 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 | mp3/mp4/mpeg/mpga/m4a/wav/webm, up to 25MB built-in speaker diarization with diarized_json output (speaker labels + segment timestamps) chunking_strategy required for audio >30s no prompt support |
| Languages | 29 GA + 82 Preview locales (111 total) across StreamingRecognize, Recognize and BatchRecognize diarization covers 14 of them | 57 languages listed for the transcriptions endpoint (one shared list for all transcription models) ISO 639-1 / 639-3 codes accepted for GPT-4o-based models |
| Max output | — | 2K |
Specs are transcribed from each vendor’s documentation; a row a vendor does not publish is left out rather than inferred. Full sources: Chirp 3 · GPT-4o Transcribe Diarize
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-3",
# model="gpt-4o-transcribe-diarize", # 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-3",
// model: "gpt-4o-transcribe-diarize", // 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-3" \
# -F model="gpt-4o-transcribe-diarize" \ # 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-3",
// Model: "gpt-4o-transcribe-diarize", // 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-3")
// .model("gpt-4o-transcribe-diarize") // uncomment this line, comment the one above
.file(Paths.get("meeting.mp3"))
.build()).asTranscription();
System.out.println(resp.text());FAQ
Which is cheaper, Chirp 3 or GPT-4o Transcribe Diarize?
They bill in different units, so there is no single honest number: Chirp 3 and GPT-4o Transcribe Diarize each appear in their own unit in the table above. Compare them on your own workload — the practical trade-off is described in the verdict at the top of this page.
Can I A/B test Chirp 3 against GPT-4o Transcribe Diarize 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 3 and GPT-4o Transcribe Diarize 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.