Fable vs. Astra
- Published
- 09/05, 08:06
- Ingested
- 09/05, 12:00
- Source type
- Dev community
- Tier
- Community
- Source status
- Healthy
Discussion trend
The percentage is based on collected discussion signal, not new comments or independent people. The curve only compares the same topic across time.
For those who use both top-tier models, what's your experience for both models?
Just wanted to test both frontier models "with same complex task and with same prompt". The fable 5.1 token consumption and ability to maintain extremely large context window is significantly cheaper and reliable than open Ai's new Gpt - 6 Astra model.
I've been using all the frontier models for many months In API platform and would dump and receive tokens of length 250k on average with max 3 turns(back and forth), since my actual work is occasional lengthy sessions, not everyday to justify pro plan. 100 or above dollars pro plan was not for me
And today, i was shocked to see the Open AI's Astra model's token hungry character and handling extremely long context window.
Tbf, i maxed out all the available variables for both models to test it to limit. And the token consumption is just day and night. The quality of output too isnt drastically better with Astra. the sheer no of citations it pulled is genuinely deranged.
My last biggest api bill was 8 dollars with fable 5. But now i had a 60 dollar bill for just 30 mmins of Astra visiting all the observable universe. Here are the pics
Sp, now do i get pro sub or just use API to finish the ongoing work? I still need atleast 8-9 calls to get the full blueprint that can be implemented by efficient, work horse models