GPT Audio 2
openai/gpt-audio-2approvedFamily gpt-audio · aliases: gpt-audio-2
Capabilities
Normalized capability surface used by hard filters and the task-signature matcher.
Modalities in
text + audio
Modalities out
text + audio
Tool use
unsupported
Structured output
unsupported
Streaming
supported
Context / max out
64K / 16K
Endpoints
Routable supply for this model — each row is a provider account + deployment region with its egress node bindings and price card.
| Endpoint | Provider | Account pool | Region | ZDR | Data collection | Prompt cache | Node bindings (P / S / E) | Price / MTok |
|---|---|---|---|---|---|---|---|---|
| ep_gptaudio_openai_use1 | OpenAI | pool-aacct_openai_pool_a_use1 | us-east-1 | ZDR | deny | no | PTier Anode_cf_use1_01STier Cnode_sg_dc_01ETier Bnode_my_kul_office_01 | $2.50 in$10.00 out$0.00 cached |
Quality by task family
Evaluation scores (0–1) feeding the router's quality term and profile quality floors.
Speech0.92
Translation0.88