GPT-5 (Product Review)
The first thing that strikes me when using OpenAI’s GPT-5 is that there isn’t a model switcher anymore. You no longer have to choose between different models — GPT-5 operates as an integrated system that’s clever about the different models it needs to apply. GPT‑5 is the new default in ChatGPT, replacing GPT‑4o, OpenAI o3, OpenAI o4-mini, GPT‑4.1, and GPT‑4.5.
This means the model can decide when it “needs to think harder or not.” There’s a smart model handling most questions, a deeper reasoning model for complex problems, and a real-time router deciding which approach to use. When I prompt it with something challenging (like the protein folding prediction problem), I’m struck by how quickly it responds with what feels like an appropriately sophisticated answer.
Testing the “Think Harder” Mode
I prompt the AI to “think really hard” about my next problem to make sure GPT-5’s deeper reasoning model kicks in. Generating a response definitely takes longer, but whilst GPT is thinking, I do have the option to go for a “quick answer” instead. I assume the chances of getting a more accurate response are higher if I let it think longer — though the trade-off between speed and accuracy will vary depending on your use case.
Generating a response definitely takes longer, but whilst the GPT is thinking, I do have to option to go for a “quick answer” instead, but I assume that the chances of getting a more accurate answer are higher if I ask it to think longer.
What this means for users
GPT-5 is making deeper reasoning models accessible to everyone (it’s free depending on usage volume), clearly trying to stay competitive with models like DeepSeek. As the ‘coding wars’ between the likes of Cursor, Replit and Claude Code heat up, OpenAI claims that “GPT‑5 is our strongest coding model to date.”
GPT-5 is making deeper reasoning models accessible to everyone (it’s free depending on usage volume), trying to stay hot on the heels of models such as DeepSeek.
As the ‘coding wars’ between the likes of Cursor, Replit and Claude Code are heating up, OpenAI claims that “GPT‑5 is our strongest coding model to date.” Indeed, all users can now use GPT-5 to build and preview apps entirely in the browser.
Main learning point: GPT-5’s integrated model approach represents a shift from user-controlled model selection to intelligent model selection, allowing for deeper reasoning model where required.
Related links for further learning:
- https://openai.com/index/introducing-gpt-5/
- https://openai.com/index/introducing-gpt-5-for-developers/
- https://every.to/vibe-check/gpt-5
- https://www.lennysnewsletter.com/p/an-exclusive-inside-look-at-gpt-5
- https://www.engadget.com/ai/gpt-5-is-here-and-its-free-for-everyone-170001066.html
- https://www.oneusefulthing.org/p/gpt-5-it-just-does-stuff
- https://www.wired.com/story/openais-gpt-5-is-here/
