Side by side
Gemini 3.5 Transcribe vs GPT Transcribe
Up to 4 models, normalised onto the same cost basis and scored through the lens you pick.
| Cost of talking | ||
| Per minute | Not conversational | $0.0045 |
| Per hour | — | $0.270 |
| 30-min interview | — | $0.135 |
| List price | Plan-basedEstimate | $0.0045/minPublished |
| Recruitment fit | ||
| Overall | 89/100 | 87/100 |
| Verdict | The pick for panel interviews and any non-English hiring — speaker attribution and Greek accuracy are both unmatched here. | The right transcription default for interviews, because interviews are recorded in noisy rooms on laptop mics. |
| Best for | Panel interviewsMultilingual hiringEvidence-linked interview notes | Interview recordingsScreening call notesTechnical role discussions |
| Watch out for |
|
|
| Capabilities | ||
| Modalities | audio | audio |
| Context window | — | — |
| Variants | Live, Recorded | Smart, Balanced, Live (streaming) |
| On HumanLike | Coworker, Clippy | Coworker, Clippy |
| HumanLike evals | ||
| Word error rate (English) | 0.06Our eval | — |
| Word error rate (Greek) | 0.03Our eval | — |
| Text still arriving after speaker stops (EN) | 666 msOur eval | — |
| Word error rate (clean audio) | — | 0.29Our eval |
| Word error rate (with additive noise) | — | 0.26Our eval |
| Median latency (9.2s clip) | — | 1390 msOur eval |
Cost per hour assumes 150 wpm and a 40% AI speaking share. Hover a figure for its full derivation.