GPT Audio Mini
Cost-efficient speech-to-speech model for voice agents, with an upgraded natural-sounding decoder.
Retired
GPT Audio Mini appears to be discontinued or shut down — End of availability July 20, 2026; end of life January 20, 2027; OpenAI recommends gpt-audio-1.5 This page is kept as an archive record and is excluded from search engines.
About
GPT Audio Mini is OpenAI's cost-efficient speech model, released January 19, 2026 alongside the full GPT Audio. It handles text and audio input and returns text or speech, with a 128K token context window, tool calling, and structured outputs. The snapshot OpenAI shipped featured an upgraded decoder for more natural voices and better voice consistency.
It was priced at $0.60 per million input tokens and $2.40 per million output tokens, roughly a quarter of the full model's rate, and was aimed at latency-sensitive voice agents. OpenAI set end of availability for July 20, 2026 and end of life for January 20, 2027, pointing users to gpt-audio-1.5, so this entry is a record of a retired generation.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
No vendor-published benchmark table; third-party voice-agent harnesses rate it a latency-efficient option at its price point
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 5/10
Scored 5/10 (verified tier, metrics-v1). Evidence: real third-party voice-agent usage through 2026, but end of availability Jul 20, 2026 and end of life Jan 20, 2027 per endoflife.date; OpenAI recommends gpt-audio-1.5. Closed weights. Launch facts corroborated across OpenRouter, Krater, and endoflife.date; docked for retirement and thin vendor benchmarking.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.0/10
- Maintenance · recency, releases, issue load
- 4.0/10
- Openness · license + docs
- 3.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · Adoption 7: cost-efficient speech model with real third-party voice-agent usage through 2026
- · Maintenance 4: released Jan 19, 2026; end of availability Jul 20, 2026 and end of life Jan 20, 2027 per endoflife.date; OpenAI recommends gpt-audio-1.5
- · Openness 3: proprietary, no weights
- · Manual 7: launch facts (Jan 19 2026; $0.60/$2.40 pricing; 128K context; upgraded decoder) corroborated across OpenRouter, Krater, and endoflife.date; docked for retirement schedule and thin vendor benchmarking