13 August 2025
Topic: Gpt 5
OpenAI brings GPT-4o back as a default for all paying ChatGPT users, Altman promises ‘plenty of notice’ if it leaves again
For now, the changes should help placate users who felt frustrated by the sudden shift to GPT-5 and deprecation of OpenAI's older LLMs....
TL;DR
OpenAI makes GPT-4o the default option for paying ChatGPT users and introduces new controls and rate limits for GPT-5
Key Takeaways:
Key Takeaways:
- Paying ChatGPT subscribers will have GPT-4o as the default option, and access to other legacy models with a new 'Show additional models' setting
- GPT-5 now has 'Auto', 'Fast', and 'Thinking' modes with 3,000 messages-per-week caps for paying subscribers
- GPT-4.5 remains exclusive to Pro users due to its high GPU cost, and GPT-4 remains available for all users
ChatGPT won’t remove old models without warning after GPT-5 backlash
After the backlash to replacing its 4o model with GPT-5, OpenAI will no longer get rid of old models without a heads up. "In retrospect, not continuin...
TL;DR
OpenAI will give users notice before retiring existing models, in response to backlash from removing GPT-4o.
Key Takeaways:
Key Takeaways:
- OpenAI has decided to keep existing models available as opt-in features for paying users.
- The company plans to provide advance notice before retiring models in the future.
- The decision is partly driven by user feedback and a desire to offer simplicity and predictability for users.
GPT-5 is a joke. Will it matter?
Article URL: https://www.bloodinthemachine.com/p/gpt-5-is-a-joke-will-it-matter Comments URL: https://news.ycombinator.com/item?id=44884107 Points: 3 ...
Scaling accounting capacity with OpenAI
Built with OpenAI o3, o3-Pro, GPT-4.1, and GPT-5, Basis’ AI agents help accounting firms save up to 30% of their time and expand capacity for advisory...
ChatGPT’s model picker is back, and it’s complicated
GPT-5 was supposed to get rid of all the complicated AI models to choose from in ChatGPT, but that's not how it's panned out....
TL;DR
OpenAI's GPT-5 rollout has been complicated, with users finding the model router not meeting expectations, and the company deciding to give users more options and flexibility instead.
Key Takeaways:
Key Takeaways:
- GPT-5 will have 'Auto', 'Fast', and 'Thinking' settings for users to choose from, giving more control over AI model performance,
- Rate limits have been increased to 3,000 messages/week with GPT-5 Thinking, and paid users can access legacy AI models, including GPT-4o, GPT-4.1, and o3, once again.
- OpenAI is recognizing the importance of per-user customization of model personality and plans to iterate on GPT-5's personality to better suit user preferences.
GPT-5 high, medium and low takes top 3 places in Livebench, OpenAI has all five models in top 5
o3, 4o, GPT-5, GPT-5-Thinking: A Case Study in Thoroughness, Accuracy, and Tone
We tested Qwen3-Coder, GPT-5 and other 30+ models on new SWE-Bench like tasks from July 2025
OpenAI is running some cheap knockoff version of GPT-5 in ChatGPT apparently
GPT-5 is now lying to your face and then pretending it didn't. OpenAl is actively enshittifying their best product.
GPT-5 Thinking has 192K Context in ChatGPT Plus
GPT-5 Style Router, but for any LLM including local.
The 1 Simple Trick That Makes Any AI 300% More Creative (Tested on GPT-5, Claude 4, and Gemini Pro)
GPT-5 after the hype
GPT-5 now listed as GPT-5-high on lmarena. A version not even accessible in ChatGPT. Promoting GPT-5 as a unified model made it look like it though. What do you think?
GPT-5 Thinking is confirmed to have 196k context window for Plus Users
Sonnet 4 (1M) just blew up the GPT-5 Death Star
GPT-5 SUCKS at creative writing