It’s a busy week in the world of AI models, as some of the top artificial intelligence companies roll out their latest updates. Just a week after Anthropic introduced version 4.7 of its Claude Opus AI model, OpenAI unveiled GPT-5.5, and China’s DeepSeek introduced a preview release of its V4 AI model.

The companies, which have all been jockeying for greater adoption and consumer and business use by extending and improving their AI model offerings, are trying to bring something new to the table as the tech industry continues to invest billions into these technologies.

The newest offerings, DeepSeek’s V4 Flash and V4 Pro AI models, are focused on reasoning and agentic AI tasks, with the Chinese company’s latest work using a Hybrid Attention Architecture. DeepSeek said this allows its models to keep query histories in mind even over a lengthy series of prompts. V4 also allows longer documents or code to be used as prompts, and the company said it has made architectural and optimization improvements. The models can also be deployed on cheaper hardware than other AI, which could lower costs for companies that want to use them.

Early last year, DeepSeek emerged seemingly out of nowhere to disrupt the established order of AI with cheaper models that rivaled those of the top ones. Since then, the company has become caught up in a backlash against Chinese AI companies; this week, the White House lashed out against Chinese AI, accusing China of widespread technology theft.

OpenAI’s latest

OpenAI, which has accused DeepSeek of using its models to create its own, this week made ChatGPT 5.5 available as part of the rollout of GPT-5.5 for paying subscribers. The focus of this latest version is coding, computer use and research, with some improvements that OpenAI said make it more intuitive to use with less human guidance. 

“It really feels like it’s setting the foundation for how we’re going to do computer work going forward, or how agent computing at scale will work,” Greg Brockman, the company’s president, told reporters in a briefing.

ChatGPT 5.5 is now available for paying ChatGPT and Codex users, along with the thinking and pro versions, with plans to bring it to the API soon. These models are built for work — specifically, coding, computer use and research.

Brockman said GPT-5.5 is more intuitive than previous models and can do more with less human guidance.

“It can look at an unclear problem and figure out just what needs to happen next,” he said.

Anthropic’s less risky AI model

Lastly, Anthropic’s Opus 4.7 is, let’s say, a more accessible and friendly version of AI than Mythos. Mythos is a highly advanced AI model that the company says is so potentially dangerous to cybersecurity that it has been kept from public release so far.

Opus has improvements that upgrade the aesthetics of its outputs, such as slide decks and documents (“tasteful” is a word the company uses to describe it). Anthropic says that this new model will take prompts more literally, rather than loosely interpreting wording, which could make it useful for vibe coders or others who need straightforward AI outputs.