Introducing GPT-6.1 Sol

OpenAI announces GPT-6.1 Sol, an upgrade to GPT-6 Sol designed to rival GPT-6 Astra on agentic coding, computer-use tasks, and professional workflows while dramatically reducing input costs. The model achieves near-Astra performance on complex benchmarks such as DeepSWE v1.1, GDP.pdf, AutomationBench, OSWorld 2.0, and Terminal-Bench Science 0.1, all at substantially lower per-task costs. Notably, GPT-6.1 Sol lowers input tokens to $0.10 per million cached input tokens and $2 per million input tokens, with output tokens priced at $10 per million, creating a cost-efficient path for building persistent agents that reuse context. OpenAI positions GPT-6.1 Sol as a balanced upgrade for everyday professional work, including writing, debugging, document understanding, and executing multi-step business workflows. Availability extends to Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex, with API access under the gpt-6.1-sol designation. An Ultrafast variant is teased for future release, promising substantially faster token generation. While Astra remains the top performer for the toughest scientific tasks, GPT-6.1 Sol offers a meaningful improvement-at-cost tradeoff for a broad set of routine to advanced tasks.