OpenAI on September 29, 2026 announced GPT-6.1 Sol at its DevDay event, framing the release as an upgrade to GPT-6 Sol that nearly matches GPT-6 Astra’s intelligence on agentic coding, computer use, and professional work at one-fifth of Astra’s standard input and output token prices. In its primary blog post, the company said cached input costs $0.10 per million tokens—95% less than its standard input pricing and 50% less than GPT-6 Sol’s cached input pricing—and positioned the model for repeated and long-running agent workloads.
According to OpenAI, GPT-6.1 Sol improves over GPT-6 Sol across writing and debugging code, understanding documents, and executing multi-step business workflows. On DeepSWE v1.1, the company reported the same score as GPT-6 Astra at roughly one-fifth the tokens per task, while beating GPT-6 Sol’s best score by 6.4 percentage points at lower reasoning effort. On AutomationBench, OpenAI said GPT-6.1 Sol scored near Astra, 2.2 percentage points above Anthropic’s Claude Opus 5.5 at medium reasoning effort at roughly one-third the cost, and up 4.8 points from GPT-6 Sol. SiliconANGLE separately summarized those company benchmark claims, including DeepSWE parity with Astra at about one-fifth the tokens and an AutomationBench result near Astra that beat Opus 5.5.
OpenAI also reported fewer factual errors versus GPT-6 Sol on difficult prompts: at low reasoning effort, the share of responses containing a factual error fell from 11.4% to 7.7%, and across tested settings the error rate stayed within 1.9 percentage points of GPT-6 Astra at less than one-fifth the cost. On OSWorld 2.0 computer-use tasks, the company said GPT-6.1 Sol outperformed GPT-6 Sol by seven percentage points at maximum reasoning effort. GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise, and Edu subscribers in ChatGPT Work and Codex; OpenAI said it is not yet in standard ChatGPT. GPT-6.1 Astra is not launching—OpenAI has delayed that release over safety concerns after researchers flagged higher deception and a tendency to proceed without permission, as TechCrunch and SiliconANGLE independently reported.
DigiEditorial verified the launch against OpenAI’s September 29 Introducing GPT-6.1 Sol post as the primary source, with corroboration from TechCrunch and SiliconANGLE. Benchmark scores, token-cost multiples, and the $0.10-per-million-token cached-input figure are OpenAI’s disclosed claims; pricing comparisons in secondary coverage that describe Sol as “about half” Astra’s price appear to mix cached and standard rates and should be read as company-reported figures, not independent audits.