Опубліковано 2026-08-16 · 赵予安
Пряма відповідь
OpenAI GPT-5 is in the NextModel catalog at 0.181 / 1.447 per million tokens, with a 400k context window. CacheSafety for this exact id is still unmeasured. Цей гайд написано для продуктових і платформних команд, які порівнюють якість моделей, вартість, політику routing і ризик rollout.
What shipped
OpenAI: GPT-5 has shipped on the NextModel gateway. The public model id is openai/gpt-5, and the public slug is openai--gpt-5. The display name is OpenAI: GPT-5, and the provider is OpenAI. The catalog entry has capabilities for streaming, tool, json, vision, and long context.
Pricing and context
The catalog lists input at $0.181 per 1M tokens and output at $1.447 per 1M tokens. The listed context window is 400000 tokens. The latency band is 1000-3000ms.
Who this is for
The listed use cases are vision, long-context, and agent. The intended audience is developers building those workflow types.
Unverified
Release date is unpublished in this catalog snapshot. Independent third-party quality scores are not in these notes. No first-party CacheSafety Bench row exists for this exact model in the NextModel bench table. Safe Hit, Bad Hit, and trap rates are not stated. No benchmark scores are listed in these notes.