I switched from Cody Pro to the Amp, but I didn’t have any gift funds to use. I paid $5 but it was gone after 10 messages. Is that how expensive it is? It does almost the same thing as Cody. Only slower, but it inserts the code for you. And because of this it has become hundreds of times more expensive? I used to pay $10 a month and was happy with Cody
Hey @arturmurwork Amp is usage based and you pay for the tokens you used up directly. For Free and Team tiers, we are not putting a margin on top of it so all costs are directly passed through.
Amp is not comparable to Cody as Amp is a step forward in the evolution of AI-assistant based programming and tool use. This product now brings the possibility for automated end-to-end tasks for the user. Cody was meant to be used as an assisted chatbot with some enhanced tools for file editing and context fetching.
I hope that clarifies your question.
Chiming in here. I have used Cody for a few months and was happy. With it being sunset, I tried Amp. I agree with arturmurwork. Pricing is way too high. 0.14 USD to analyse a 313 line javascript file and answer 1 question is not sustainable. Escpecially since a simpler model would do my tasks easily and my team and I have no need for agentic software just yet.
I am forced to look for something else now.
Just some feedback. After looking around and noticing 2 things, I decided to take a fresh perspective on what could justify the cost :
-
I am surprised by the speed of results. It is much faster than the other AI coders I have tried so far (cline, roo).
-
By referencing specific lines instead of ingesting the whole file, it still gets the job done with lower tokens. We develop on drupal so the code structure and syntax is already known to the LLM. No need to chew up context tokens with unneccesary references and inclusions
I have decided to use it for 30 days and take the hit on the bill while comparing speed of task completion to a fellow coder [who uses another toolset]. I will post my findings thereafter.
I have to agree
- Lines of Code +154 -52 ~21
- Threads 1
- Messages 17
0.73 USD spent, I even don’t know it is working or not… that’s a bit too expensive to afford
Thank you for taking a second chance. It is indeed another mental model to construct coming from Cody and using Amp in a more semi-autonomous end-to-end workflow experience.
We are glad for further feedback. Please join us at the Amp Discord for RT talk.
Thank you