OpenAI Unveils GPT-5.6 Model Family and Launches Ultrafast Mode Preview
모델 출시 | Sat Aug 15 2026 00:00:00 GMT+0000 (Coordinated Universal Time) | 3 sources
The GPT-5.6 series improves agent performance and cost efficiency, while the new Ultrafast mode delivers speeds of 750 tokens per second.
Analysis
[OpenAI] unveiled the GPT-5.6 model family [1]
- Significantly improved agent (autonomous AI) performance and price efficiency
- Lineup consists of three models: Sol
- Luna
- and Terra
- Higher accuracy than the previous generation even at low reasoning effort
[OpenAI] launched Ultrafast mode preview [2]
- Runs GPT-5.6 Sol up to 14x faster than standard
- Generates up to 750 output tokens per second
- Available first on the OpenAI API
[OpenAI × Cerebras] revealed Ultrafast infrastructure partnership [3]
- Leverages high-speed inference infrastructure based on Cerebras chips
- Currently operating in preview for a small number of customers
- Access will gradually expand as capacity grows
[Responses API] added new default features for agent efficiency [1]
- Supports reasoning persistence across model turns
- Introduces native compaction to compress long conversations
- Enhanced multi-agent orchestration and programmatic tool calling
[GPT-5.6 Sol Ultrafast] presented real-time business use case scenarios [2]
- Real-time analysis of logs and code changes during incident response
- Real-time detection of financial market signals and suspicious transactions
- Solves complex problems without interrupting customer support and voice conversations
- Prevents commerce checkout abandonment and enables live research experiments