GPT-4o Mini Transcribe (OpenAI)

GPT-4o Mini Transcribe is OpenAI's compact speech-to-text model in the GPT family, designed to convert spoken audio into written text. It sits alongside its sibling GPT-4o Transcribe as a lighter-weight ASR option aimed at workloads where lower cost and faster throughput matter more than the absolute lowest error rates. Independent benchmark coverage places it on Artificial Analysis's non-streaming WER leaderboard alongside other proprietary transcription systems, giving developers a third-party view of where it lands among competing speech engines.

In practical terms, the model is best suited to moderate-length transcription tasks where its efficiency advantages outweigh a slightly higher word error rate than top-tier alternatives. Evaluators at Artificial Analysis had to chunk long Earnings22 audio to 9-minute segments because the model could not reliably process full-length recordings under its default time limits, so very long files need to be split before submission. OpenAI has also announced a planned shutdown for this model, with gpt-live-transcribe or gpt-transcribe named as successor options, so teams building on it should plan a migration path rather than treating it as a long-term production choice.

LLM Gatewayopenai/gpt-4o-mini-transcribegpt

Open provider page Provider docs

Quick Info

Provider: LLM Gateway
Model key: openai/gpt-4o-mini-transcribe
Release date: Mar 20, 2025
Last updated: Mar 20, 2025
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $1.25
Output token cost: $5.00

Limits

Output tokens: 2,000 tokens
Context window: 16,000 tokens

Transparent token rates

Compare GPT-4o Mini Transcribe (OpenAI) pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about GPT-4o Mini Transcribe (OpenAI)

LLM Gateway

Coverage

AI Flash Report's Week 36, 2026 tracker alert (covering August 31 to September 6, 2026) records an announced shutdown for gpt-4o-mini-transcribe (OpenAI) scheduled for February 26, 2027, with replacement models identified as gpt-live-transcribe or gpt-transcribe. The entry is part of a seven-item deprecation batch in t For developers and integrators using GPT-4o Mini Transcribe via any routing layer, this signals roughly five and a half months of runway before the model goes offline, with named successor SKUs (gpt-live-transcribe and gpt-transcribe) to plan migrations toward. The data is sourced from a third-party tracker rather than

LLM Gateway

CoverageBenchmark

Artificial Analysis's non-streaming Speech-to-Text leaderboard explicitly lists "GPT-4o Mini Transcribe, OpenAI" among 36 of 71 proprietary and open-weights models evaluated on the AA-WER v2 Word Error Rate Index, with companion axes for speed (audio seconds transcribed per second) and price (USD per 1000 minutes of au For developers this surfaces a concrete, model-specific operational constraint: long-form Earnings22-style audio must be segmented to 9-minute chunks before transcription by GPT-4o Mini Transcribe, or accuracy evaluations against that benchmark will not reflect true end-to-end behavior. The leaderboard frames the model