In previous months we mostly saw new model releases. July points to a different trend: AI systems are profiling themselves more as tools for specific tasks. Agents, study modes, more autonomy.
This month, along with the already confirmed news, we've gathered a few rumors currently going around the AI world. Whether they actually happen is still an open question.
OpenAI Agent
A new tool from ChatGPT that can autonomously perform actions across connectors.
- The agent has its own workspace that you open directly in the chat.
- For Pro subscribers and above, it has access to all the connectors you've connected, for example Google Drive, Calendar, or GitHub.
- It can handle specific actions: create a calendar event, order goods, analyze documents, and so on.
- It is semi-autonomous.
- The agent's structure was inspired by a model that placed very well in the International Mathematical Olympiad (IMO).
- It's actually an iteration of OpenAI Operator, which worked similarly but with a much worse UX.
Useful links: Introducing ChatGPT Agent, Connector list, List of possible actions
ChatGPT Study Mode
Alongside the agent, OpenAI also introduced a new Study Mode, which you can turn on right in GPT chat.
- ChatGPT takes on the role of a teacher or tutor.
- You go through the material step by step, at a pace that suits you.
- The GPT doesn't hand you the answer right away. It tries to guide you and pull you into thinking, and it can also test you on the material.
Useful links: ChatGPT Study Mode
Grok 4 and Grok 4 Heavy
Elon Musk's xAI released new versions of its model in July, Grok 4 and Grok 4 Heavy.
- Both versions are available only in the paid plan.
- According to published data, they reach the highest scores in the so-called Intelligence Benchmark Index (Artificial Analysis).
Useful links: News Grok 4
Gemini 2.5 Deep Think
Google extended its Gemini model line.
- The version aims at better reasoning and multi-step planning.
- Deep Think should enable agents to run in parallel, that is, scenarios where several separate processes run at once.
- Available only to Ultra subscribers.
Useful links: GoogleAI Status
Extremely fast models from Cerebras
Cerebras hosts extremely fast language models that are well ahead of the competition in processing speed.
- The platform can host very large models.
- But it's also one of the most expensive options.
- In generation speed and response, it's at the very top.
- Despite the high cost, it can be an interesting choice for companies that tackle demanding tasks.
Useful links: Artificial Analysis Status
And what's no longer confirmed news, but still worth mentioning:
Open-source model from OpenAI
From several hints and leaks, it looks like OpenAI is working on an open-source release. The details aren't confirmed yet, but:
- There's talk of a smaller language model that could be an alternative to open-source models like LLaMA or Mistral.
- The release could be a response to pressure from the community and competitors. Meta, Mistral, and Cohere already offer open models.
ChatGPT 5 during autumn
- OpenAI hasn't officially announced anything yet, but has hinted several times that a "major update is near".
- Improvements in reliability, reasoning abilities, and long-term memory are expected.
Meta offers astronomical sums to AI engineers
- You can reportedly get up to $1.25 million for a four-year contract. That works out to some $312,000 a year.
- Even so, many candidates turn it down. These are mainly employees of competing companies such as OpenAI.
Useful links: Tom's Hardware Article
July showed a shift from new models to their real use.
AI is becoming more practical, more autonomous, and closer to everyday scenarios.
Want to stay one step ahead?
Don't miss our best insights. No spam, just practical analyses, invitations to exclusive events, and podcast summaries delivered straight to your inbox.