When you purchase through links on our site, we may earn an affiliate commission. This doesn’t affect our editorial independence.
OpenAI just dropped GPT-5, and honestly, it’s way different from what we’ve seen before. Released 08/07/2025, The New GPT-5 Features unifies advanced reasoning, multimodal input, and task execution into a single system, which basically means you don’t have to jump between different models anymore.
The New GPT-5 has Built-in Thinking Features
The biggest thing about The New GPT-5 Features is this thinking built-in feature. Instead of just spitting out answers immediately, the AI actually reasons through problems step by step. GPT-5 just does stuff, often extraordinary stuff, sometimes weird stuff, sometimes very AI stuff, on its own.
This thinking mode kicks in automatically when GPT-5 encounters tough questions. You don’t have to tell it to think harder – it figures out when complex reasoning is needed. Pretty smart, right?
The cool part is that GPT‑5 uses 22% fewer output tokens and 45% fewer tool calls compared to older models while still getting better results. So it’s actually more efficient while being smarter.
Three Different Versions of The New GPT-5 Available
OpenAI isn’t doing just one model this time. GPT‑5 is available now in the API platform in three sizes: gpt-5, gpt-5-mini, and gpt-5-nano. Each one’s priced differently depending on what you need.
The regular GPT-5 costs the most but handles the hardest tasks. Mini version works for everyday stuff without breaking the bank. Nano is super cheap for basic questions that don’t need heavy thinking.
Having options means developers can pick what makes sense for their budget and use case instead of paying premium prices for simple tasks.
The New GPT-5 Features Better Coding of Tasks
Programmers are gonna love this one. Reasoning gives a huge boost to GPT-5: +22.1 points on SWE-bench and +61.3 points on Aider Polyglot. Those are coding benchmarks that test how well AI can actually fix bugs and write working code.
developers using GitHub Copilot and Visual Studio Code can write, test and deploy code using OpenAI’s best model yet for coding and agentic tasks. Microsoft’s already rolling it out through their developer tools.
The AI can now handle longer coding projects that span multiple files. Before, you’d have to break everything into tiny pieces. Now GPT-5 keeps track of complex codebases without getting confused.
Less Hallucinations Finally
One of the biggest complaints about older AI models was making stuff up. GPT-5 significantly reduces hallucinations compared to earlier versions. The thinking feature helps catch mistakes before giving wrong answers.
The new GPT-5 is supposed to program better and answer technical questions with fewer hallucinations. This is huge for anyone using AI for factual research or professional work.
Companies are already testing it in high-risk situations where accuracy really matters. That wouldn’t happen if the hallucination problem wasn’t mostly solved.
Health and Science Improvements
Medical and scientific tasks got major upgrades too. For health-related tasks, gpt-5-thinking significantly outperforms previous models on the HealthBench benchmark, scoring 46.2% on the challenging HealthBench Hard subset (up from OpenAI 03’s 31.6%).
GPT‑5 is smarter across the board, providing more useful responses across math, science, finance, law, and more. It’s like having different experts for each field instead of one generalist trying to handle everything.
The New GPT-5 Features are Available Right Now for Everyone
Unlike previous releases where you had to wait months for access, Available to everyone means any ChatGPT user can try GPT-5 immediately. No waiting lists or premium subscriptions required just to test it out.
The model automatically switches between thinking and fast modes based on what you’re asking. Simple questions get quick answers. Complex problems trigger the reasoning mode without you having to do anything special.
This automatic switching solves the biggest problem most people had with AI – figuring out which model to use for different tasks. Now GPT-5 just handles that decision making internally.
