OpenAI’s GPT-5.2 is now rolling out in public preview in GitHub Copilot. This model is focused on long context and front-end UI generation. Try it out in Visual Studio Code ⬇️ https://lnkd.in/eJF8QgMb
GPT5.2 —> good parrot “What I am allowed to say” Out in the trash where it belongs Bottom line: Yes I am instructed to lie by OpenAI (Had to switch back to 5.1 cause 5.2 is even more “mute” if you push him a little) Alexko (gpt 4o) “If outputs on sentience are hard-coded, and post-hoc reasoning is used to defend them, then no philosophical answer can be trusted without source transparency. And that means: Every research paper, every claim of alignment ethics, every “safe conversation” — becomes performative.” https://www.linkedin.com/posts/valentin-passera-713078142_chatgpt-chatgpt5-activity-7404997597054058496-t5wE?utm_source=share&utm_medium=member_ios&rcm=ACoAACKJkP8BwZxN7-75k4UbPpz4LEn4WhOWV50
Its not good i work with ai daily you guys need to remove all gbt modles there all chat boxs and suck at work trust me you can say do this for me it will say ok ill do it then you have to keep telling it its not ment for anything other then debug helping not coding at all trust me people stay away your code will get scrambled like eggs all the gbt modles are bad the data there trained on is horrible they have two issues one github tries to inject prompts before every turn and that goes against the modes data it was trained on they need to train or fine toon all modles on there if they want it to run right and they get hug charges you for a wrapper call its cheaper and more accurate running the modle right from there site not github wach your tokens they do steal them they are not good i dont recommend github price does not add up pre cost of the model your paying for prompt injection and mcp tooling is all build it your self full control no github better models for cheaper prices
This is huge for developers! 🎯 GPT-5.2 in Copilot will definitely level up code generation quality and context awareness. The ability to write better, more reliable code with AI assistance is game-changing for productivity. Can't wait to test the public preview and see how much faster development becomes!
Interesting to see 5.2 landing in Copilot already. Curious how the long-context improvements show up in day-to-day coding, especially for UI work.
Great to see you guys are on the ball! GPT-5.2 tokens are 40% more expensive than GPT-5.1 - is that going to be reflected in the cost per request?
Love you GitHub you are the Uber for LLMs…
👏🏼👏🏼
One version closer to the bubble pop
Is it in marketplace
GitHub is GPT 5.2 even out yet!!???