Back
PRATHAM
Experimenting On lea... • 1d
After GPT-5 dropped, the internet lost its mind. Suddenly, every tech influencer was screaming, “Front-end jobs are over!”, “ChatGPT has finally launched GPT-5!”, “It’s AGI!”, “It’s everything AGI!” all because it says so on the website. Stop falling for people who just skim a landing page, copy-paste the hype, and never bother to actually test or look at the numbers. It’s a model that can automatically switch its “brain” depending on your question that’s the main upgrade. You don’t have to manually pick a reasoning model, a creative model, or a code model anymore; it figures it out on its own. Useful? Sure. Revolutionary? No. If you think GPT-5 can “take your job,” then Claude Opus 4 was already capable of similar reasoning months ago. In fact, GPT-5’s thinking ability is basically on par with Opus 4. Which means, if your job was at risk, that threat has been around for a while now. SWE-bench Verified (real-world coding): GPT-5 scores 74.9%, Claude Opus 4.1 scores 74.5% — that’s a 0.4% difference, not a revolution. Aider Polyglot (multi-language coding): GPT-5 hits 88%, cutting errors by one-third compared to o3, but Claude hasn’t published this number.
Replies (4)
More like this
Recommendations from Medial
Aakash kashyap
Building JalSeva and... • 9m
AI Model Performance Benchmarks: 🚀 Comparing Claude, GPT-4o, and Gemini Across Key Tasks with Claude 3.5 Sonnet performing best overall, especially in code (93.7%) and reasoning (65.0%). Gemini 1.5 Pro excels in math (86.5% with 4-shot CoT). GPT-
See More
Account Deleted
Hey I am on Medial • 1y
Hello Guys 🤩, Claude Opus model is dominating the AI race with 86.8% by competing with GPT-4 with 256 K token processing power and GPT-4 with 138 K tokens . Rumor Is that GPT-5 will launch in August with 1 Million tokens , isn't it interesting 😲
See More
Chamarti Sreekar
Fcuk imposter syndro... • 5m
Sam Altman just announced key updates to OpenAI’s roadmap: The free tier of ChatGPT will have unlimited access to GPT-5 at the standard intelligence level. GPT-4.5 (codenamed Orion) will be the last model without chain-of-thought reasoning. OpenAI
See More
Download the medial app to read full posts, comements and news.