GPT-4o Mini Transcribe vs GPT-4o Transcribe: price & specs, measured

Which one, when

The mini tier roughly halves the bill - $1.25 per million text input, $5 output and $3 per million audio input against $2.5, $10 and $6 - for the same 25MB file cap, the same 57-language list, the same 16000-token context and the same json/text-only output with no timestamps or diarization on either. Nothing else in the published spec separates them, so pick gpt-4o-mini-transcribe unless you have measured a difference on your own audio.

Benchmarks

GPT-4o Mini Transcribe: 33 published, but no benchmark it shares with enough other models to compare.

Above averageGPT-4o Transcribeonly 1 comparable

GPT-4o Mini Transcribe GPT-4o Transcribe other models measured peer average ★ no peer scored higher

FLEURS Arabic (WER)

14.06%

11.3%

FLEURS Arabic (WER)

14.06%

11.3%

FLEURS Bengali (WER)

24.45%

18.51%

FLEURS Danish (WER)

8.72%

6.14%

FLEURS Dutch (WER)

5.44%

3.78%

FLEURS English (WER)

2.86%

2.46%

FLEURS French (WER)

4.65%

3.46%

FLEURS German (WER)

3.47%

2.61%

FLEURS Gujarati (WER)

33.33%

24.98%

FLEURS Hindi (WER)

12.91%

11.14%

FLEURS Indonesian (WER)

4.77%

3.44%

FLEURS Italian (WER)

2.08%

1.63%

FLEURS Japanese (WER)

4.05%

3.06%

FLEURS Kannada (WER)

34.36%

27.46%

FLEURS Korean (WER)

4.5%

4.07%

FLEURS Malay (WER)

7.01%

4.21%

FLEURS Malayalam (WER)

44.09%

32.62%

FLEURS Mandarin Chinese (WER)

7.71%

7.03%

FLEURS Marathi (WER)

29.62%

23.1%

FLEURS Norwegian Bokmal (WER)

7.06%

5.05%

FLEURS Polish (WER)

4.75%

3.25%

FLEURS Portuguese (WER)

3.36%

2.63%

FLEURS Romanian (WER)

5.78%

4.08%

FLEURS Russian (WER)

4.34%

3.54%

FLEURS Spanish (WER)

2.44%

1.96%

FLEURS Swahili (WER)

25.31%

13.85%

FLEURS Swedish (WER)

7.1%

4.87%

FLEURS Tamil (WER)

42.79%

33.64%

FLEURS Telugu (WER)

33.47%

29.15%

FLEURS Thai (WER)

6.13%

4.58%

FLEURS Turkish (WER)

5.53%

4.42%

FLEURS Ukrainian (WER)

5.72%

4.01%

FLEURS Urdu (WER)

19.25%

14.25%

FLEURS Vietnamese (WER)

5.68%

3.76%

WenetSpeech test-net (CER)

N/A

15.3%

FLEURS 53-lang (WER)

N/A

8.17%

AISHELL-2 test (CER)

N/A

4.24%

CommonVoice en (WER)

N/A

9.08%

CommonVoice yue (CER)

N/A

11.36%

CommonVoice zh (CER)

N/A

6.32%

Dialog-Accented English (WER)

N/A

28.56%

Dialog-Cantonese (CER)

N/A

16.05%

Dialog-Chinese Dialects (CER)

N/A

45.37%

Dialog-Mandarin (CER)

N/A

20.73%

Elders&Kids (CER)

N/A

14.27%

EntireSongs-en (WER)

N/A

30.71%

EntireSongs-zh (CER)

N/A

34.86%

ExtremeNoise (CER)

N/A

36.11%

Fleurs-yue (CER)

N/A

4.98%

GigaSpeech (WER)

N/A

25.5%

KeSpeech (CER)

N/A

26.87%

Librispeech test-other (WER)

N/A

3.75%

M4Singer (CER)

N/A

16.77%

MIR-1K vocal (CER)

N/A

11.87%

Opencpop (CER)

N/A

7.93%

PopCS (CER)

N/A

32.84%

SpeechIO (CER)

N/A

12.86%

TED-LIUM (WER)

N/A

7.69%

TongueTwister (CER)

N/A

20.87%

VoxPopuli (WER)

N/A

10.29%

WenetSpeech test-meeting (CER)

N/A

32.27%

WenetSpeech-yue long (CER)

N/A

25.29%

WenetSpeech-yue short (CER)

N/A

15.62%

Vendor-published: Alibaba (Qwen) ByteDance Google OpenAI

Pricing

	GPT-4o Mini Transcribe	GPT-4o Transcribe	Δ
Audio input / 1M tokens	$3	$6	0.5×
Text output / 1M tokens	$5	$10	0.5×

Rates from the live catalogue at build time; each model page carries the current rate card.

Capabilities

	GPT-4o Mini Transcribe	GPT-4o Transcribe
Speaker diarization	no	no
Streaming	yes	yes
Timestamps	no	no

Specs

	GPT-4o Mini Transcribe	GPT-4o Transcribe
Input modalities	text audio	text audio
Output modalities	text	text
Released	2025-03-20	2025-03-20
Knowledge cutoff	2024-06	2024-06
Limits	mp3/mp4/mpeg/mpga/m4a/wav/webm, up to 25MB streaming transcription supported json/text output only no word timestamps or diarization	mp3/mp4/mpeg/mpga/m4a/wav/webm, up to 25MB streaming transcription supported (incl. Realtime transcription sessions) json/text output only no word timestamps or diarization
Languages	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	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	2K

Specs are transcribed from each vendor’s documentation; a row a vendor does not publish is left out rather than inferred. Full sources: GPT-4o Mini Transcribe · GPT-4o Transcribe

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="gpt-4o-mini-transcribe",
    # model="gpt-4o-transcribe",  # 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: "gpt-4o-mini-transcribe",
  // model: "gpt-4o-transcribe",  // 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="gpt-4o-mini-transcribe" \
  # -F model="gpt-4o-transcribe" \  # uncomment this line, comment the one above
  -F file=@meeting.mp3

package 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: "gpt-4o-mini-transcribe",
		// Model: "gpt-4o-transcribe",  // 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("gpt-4o-mini-transcribe")
        // .model("gpt-4o-transcribe")  // uncomment this line, comment the one above
        .file(Paths.get("meeting.mp3"))
        .build()).asTranscription();
System.out.println(resp.text());

Get your API key →

FAQ

Which is cheaper, GPT-4o Mini Transcribe or GPT-4o Transcribe?

GPT-4o Mini Transcribe is cheaper on the "Audio input / 1M tokens" row ($3 vs $6, 2.0× apart). Other rows may point the other way; the table above carries the full rate card, and real cost depends on your mix.

Can I A/B test GPT-4o Mini Transcribe against GPT-4o Transcribe without two integrations?

Yes. Both are served through the same OpenAI-compatible endpoint with one API key. Switching is a one-line change to the model id, so you can route a fraction of traffic to each and compare bills directly.

Do GPT-4o Mini Transcribe and GPT-4o Transcribe 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.

Related comparisons

GPT-4o Transcribe vs GPT-4o Transcribe Diarize GPT-4o Transcribe vs Whisper v1 Compare any two models →

From our measured studies

Voice Agent API Cost: a 10-Minute Call Runs $0.04 to $0.57 Speech-to-Text APIs: 14 Models, $0.002 to $0.016 a Minute