OpenAI to Release GPT-5.4 with 2 Million Token Context Window

According to information circulating online, a new version of OpenAI's language model, GPT-5.4, is in development. References to it were found in public pull requests in the Codex repository, but these were subsequently removed.

OpenAI to Release GPT-5.4 with 2 Million Token Context Window

Sources report that a key innovation could be an expanded context window of up to 2 million tokens. This will allow the model to process large data sets, such as large codebases or long document collections, without the need to re-enter context.

Additionally, support for full processing of PNG, JPEG, and WebP images is expected. This could be useful for analyzing technical schematics, interface screenshots, and documents with small text, where compression typically results in the loss of important details.

Other potential updates include the introduction of a high-speed tariff and a so-called "extreme" reasoning mode, which will allow the model to allocate more computing resources to solving complex multi-stage problems, which could be useful in research. Prediction markets estimate the probability of the model's release by April 2026 at 55%, and by June at 74%.

The Codex app for Windows, previously only available on macOS, has also been announced. It's available in the Microsoft Store for ChatGPT users without a subscription, as well as with Go, Plus, and Pro subscriptions. It allows you to work with multiple agents simultaneously, perform code reviews, and use ready-made solutions from GitHub.

Share

What's Your Reaction?

Like Like 0
Dislike Dislike 0
Love Love 0
Funny Funny 0
Angry Angry 0
Sad Sad 0
Wow Wow 0