However, other users find value in the service, particularly the 3.5-turbo version, which they find cost-effective and useful for intelligent decision making in their applications and backend processes. Some users also find the 4 version useful for their custom GPTs, despite the forced 1.5k token system prompt in every response.
Key takeaways:
- Some users are considering cancelling their ChatGPT subscription due to its recent instability and failure issues.
- Others argue that GPT 3.5-turbo is a cost-effective and intelligent alternative to GPT 4, despite being a bit slower.
- However, some users find GPT 3.5 unreliable and less useful for detailed prompts, preferring GPT 4 for their needs.
- There is criticism over the new 1.5k token system prompt being forced into every response, making the API more sensible for most purposes.