Solar Pro 4 is the agentic model that finishes the job. Give it a contract, a spreadsheet, or a terminal task, and it reads the documents, runs the tools, and produces the deliverable — then stops and says so when the evidence runs out.
Built for agent work
Solar Pro 4 performs best on the evaluations that decide whether real work gets done:
- Terminal tasks (Terminal-Bench v2.1): 57
- Multi-turn tool use (τ³-Banking): 23
- Long-document reasoning (AA-LCR): 71
Each of these is a several-fold step over Solar Pro 3.
Trained on finished work
Solar Pro 4 was trained and validated on office work in the same shape it takes in the real world. OfficeVerse synthesizes tasks from real public data across 11 industry domains and 12 task types, grading each pass or fail on the final deliverable.
Answers on evidence
Solar Pro 4 is built to say it cannot verify rather than fill the gap. Find what is there — don't invent what isn't.
Start chatting
Sign in with Google and start a chat on the home page. New accounts start with free credits.
