GPT-5.6 Sol is OpenAI's new flagship model for ambitious agentic work. It leads its own family on coding, security and medical questions. On TerminalBench 2.1 it scores 88.8 percent. Sol Ultra tops the same test at 91.9 percent.
What is GPT-5.6 Sol?
GPT-5.6 Sol sits at the top of OpenAI's GPT-5.6 line. Named after the Sun, it offers the highest reasoning effort in the series. The model targets long, multi-step agent tasks. Building a software project end to end is a good example.
The GPT-5.6 family: Sol, Terra and Luna
OpenAI ships GPT-5.6 in three models. Each one sits on a different budget and speed balance.
- Sol: flagship model for hard agent work.
- Terra: balanced option for everyday work.
- Luna: fast, affordable model for high-volume work.
How much does GPT-5.6 Sol cost?
| Model | Input | Cached Input | Output |
|---|---|---|---|
| GPT-5.6 Sol | $5.00 | $0.50 | $30.00 |
| GPT-5.6 Terra | $2.50 | $0.25 | $15.00 |
| GPT-5.6 Luna | $1.00 | $0.10 | $6.00 |
What do the benchmarks show?
TerminalBench 2.1
TerminalBench measures how often a model finishes real terminal tasks. GPT-5.6 Sol Ultra leads at 91.9 percent, with standard Sol at 88.8 percent.
HealthBench Professional
On HealthBench, Sol reaches about 60.5 percent and leads the GPT line on the score to price frontier.
ExploitBench
On ExploitBench, the Sol curve reaches the 73 percent band, well above Terra and Luna.
Who is GPT-5.6 Sol for?
Sol makes sense for teams that need the highest success rate on complex code bases and long agent tasks. For simple work Terra is often enough, while Luna stays cheaper for high volumes.
Read the full details on OpenAI's GPT-5.6 announcement and the Sol preview page.