Description
Describe the bug
When connections are interrupted due to unstable network, you need to resend the code multiple times, wasting many tokens.
Often, even though the code preview panel on the right successfully generates the code, only half of it is displayed in the end. For example, if 200 lines are generated, it only shows 50 or 100 lines. This happened multiple times with no solution.
Due to issue #2, I can only have it output complete code as text in the chat dialog, but then I frequently encounter issue #1, leading to endless repetition. Amazingly, I used up all tokens in just a few hours.
Theoretically, with the right side code displaying normally, a complete webpage generation shouldn't use more than 20% of tokens. Yet due to issues 1, 2, and 3 occurring simultaneously with 2 JS and Python code files (each less than 300 lines), I burned through 9,888,360/10,000,000 tokens in one evening.
Link to the Bolt URL that caused the error
Steps to reproduce
When connections are interrupted due to unstable network, you need to resend the code multiple times, wasting many tokens.
Often, even though the code preview panel on the right successfully generates the code, only half of it is displayed in the end. For example, if 200 lines are generated, it only shows 50 or 100 lines. This happened multiple times with no solution.
Due to issue #2, I can only have it output complete code as text in the chat dialog, but then I frequently encounter issue #1, leading to endless repetition. Amazingly, I used up all tokens in just a few hours.
Theoretically, with the right side code displaying normally, a complete webpage generation shouldn't use more than 20% of tokens. Yet due to issues 1, 2, and 3 occurring simultaneously with 2 JS and Python code files (each less than 300 lines), I burned through 9,888,360/10,000,000 tokens in one evening.
Expected behavior
When connections are interrupted due to unstable network, you need to resend the code multiple times, wasting many tokens.
Often, even though the code preview panel on the right successfully generates the code, only half of it is displayed in the end. For example, if 200 lines are generated, it only shows 50 or 100 lines. This happened multiple times with no solution.
Due to issue #2, I can only have it output complete code as text in the chat dialog, but then I frequently encounter issue #1, leading to endless repetition. Amazingly, I used up all tokens in just a few hours.
Theoretically, with the right side code displaying normally, a complete webpage generation shouldn't use more than 20% of tokens. Yet due to issues 1, 2, and 3 occurring simultaneously with 2 JS and Python code files (each less than 300 lines), I burned through 9,888,360/10,000,000 tokens in one evening.
Screen Recording / Screenshot
No response
Platform
Browser name = Safari
Full version = 17.5
Major version = 17
navigator.appName = Netscape
navigator.userAgent = Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.5 Safari/605.1.15
performance.memory = undefined
Username = nico0766
Chat ID = ffb491ea4093
Additional context
No response
Activity