Mistral 7B is 187x cheaper compared to GPT-4
medium.com
Mistral 7B is 187x cheaper compared to GPT-4
Introducing LoraLand: 25 Fine-Tuned Mistral-7B Models That Outperform GPT-4 š
Predibase has launched LoraLand š, which consists of 25 fine-tuned Mistral-7B models surpassing GPT-4 in task-specific applications, ranging from sentiment detection to question answering.
Predibase claims that a... See more
sanity check on @cognition_labs
Per task, Devin is likely doing 10-20 maxed out GPT4-32k calls per minute, thats ~$2-$5/min, or $120-$300/hr depending on Input/Output token ratio.
my fellow waterloo interns will gladly outperform Devin for that hourly cost :)
brian-machado-high-inferencex.com
šØ š¢ Releasing BARGAIN: Guaranteed Accurate AI for Less
š° BARGAIN reduces LLM powered data processing costs by using cheaper LLMs (e.g,gpt-4o-mini) on data records they are accurate on
šÆ It *theoretically guarantees* the output matches expensive LLM's output (e.g,gpt-4o) https://t.co/SN7uWD0NIp