ChatGPT's GPT-5 might be just a few days or weeks away, as we've spotted references to a new model called gpt-5-reasoning-alpha-2025-07-13 . As spotted on X, OpenAI is testing a model called "gpt-5-reasoning-alpha-2025-07-13" . This model was finalised on the 13th of July, and it appears to be the final round of testing .
"Models: openai/gpt-5-reasoning-alpha-2025-07-13: reasoning_effort: high," one of the code references read, suggesting that it focuses on complex reasoning tasks . Alexander Wei, a researcher at OpenAI, recently confirmed that GPT-5 is on its way . Other researchers have also dropped hints that GPT-5 will combine breakthroughs from all models to create a unified model .
GPT-5 is expected to bring significant improvements in reasoning ability, reduced hallucinations, and improved context understanding . OpenAI CEO Sam Altman said that GPT-5 will be released only when it is safe and ready . It is also expected to natively support text, image, audio, and voice, potentially including video, through a unified interface .
Furthermore, GPT-5 will have a significantly larger memory and context window, allowing it to process massive documents and lengthy conversations without losing context . It is also expected to be more capable of acting autonomously, performing tasks end-to-end and learning from interactions .
OpenAI is also testing o3-alpha, which is better than o3 and o3-pro in coding and front-end designing . OpenAI models have always performed poorly when it comes to frontend designing, but that might change with o3-alpha . It's also possible that GPT-5 will ship with o3-alpha capabilities we spotted earlier .
While OpenAI stays quiet on the release of GPT-5, the detectives of the internet are certain it will be launching in July, or at the very latest this summer .