OpenAI names Prabhjeet Singh India MD, launches GPT-5.6 Sol with stronger cyber protections

OpenAI has named former Uber India head Prabhjeet Singh as the Managing Director for India. Meanwhile, the ChatGPT maker has also announced the limited release of its latest GPT 5.6 model.

OpenAI CEO Sam Altman confirmed reports that GPT 5.6 will see a staggered launch. Altman said that the new models will be released in limited preview and that the company is working with the US government to get to “general availability as fast as we can.”

“We don’t believe this kind of government access process should become the long-term default. It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them,” OpenAI wrote in a blog post.

What’s new with GPT 5.6?

OpenAI says GPT 5.6 is its most capable AI model family yet and has introduced three variants: Sol, Terra and Luna. The Sol model is the flagship variant, while Terra is a mid-tier model and Luna is the base variant aimed at users looking for faster responses without needing a subscription.

OpenAI says that Terra offers similar performance to GPT 5.5 while being two times cheaper.

In a post on X, Altman said that the new naming scheme was heavily inspired by the answers generated by the GPT 5.5 model.

The company calls GPT 5.6 Sol its “strongest model yet” and has also added a new Max Reasoning setting that allows the AI to reason more deeply. OpenAI is also bundling an Ultra mode with its new AI, which will allow it to leverage sub-agents in order to accomplish complex work faster.

In terms of benchmarks shared by OpenAI, the company says GPT 5.6 Sol and Sol Ultra both beat Claude Mythos 5 in the Terminal-Bench 2.1 benchmark, which measures coding performance through planning, iteration and tool coordination.

On GeneBench v1, GPT 5.6 Sol outperformed GPT 5.5 on genomics and quantitative biology tasks while using fewer output tokens.

OpenAI also calls Sol its strongest cyber model so far and says it offers improved performance on long-horizon vulnerability research and exploit development-related tasks.

On ExploitBench, OpenAI says GPT 5.6 Sol achieved performance close to Anthropic’s Mythos Preview while using roughly one-third of the output tokens.

The company also announced that GPT 5.6 is its model with the most robust safeguards to date. OpenAI says it has added multiple layers of safeguards for higher-risk activity, sensitive cyber requests and repeated misuse. The company also says it has spent weeks finding weaknesses, pressure-testing its systems and hardening them against real-world attacks.

“Our goal is to make prohibited offensive activity more difficult, uncertain, and detectable without unnecessarily limiting those beneficial uses,” OpenAI said in its blog.

“GPT 5.6 Sol is better at helping people find and fix vulnerabilities than reliably carrying out end-to-end attacks,” it added.

OpenAI appoints Prabhjeet Singh as India MD

OpenAI has announced that Prabhjeet Singh will lead the company’s India operations and oversee areas such as consumer growth, enterprise adoption and strategic partnerships in the country.