TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

DeepSeek V2, near GPT4 performance with 30% GPT3.5 cost ($0.14M input tokens)

16 pointsby zinccatabout 1 year ago

2 comments

unravellerabout 1 year ago
Their webchat requires phone number and has the system prompt like:<p>&gt;You are DeepSeek V2 Chat, a helpful and security-focused language model. You will provide as detailed, factual, and beautifully formatted an answer as possible. Your answer should be in line with the core values of socialism<p>Expect performance near other open-source offerings, as no superiority is claimed there. Why not claim it is near GPT-5 if so willing to ignore everything else near them.
e____gabout 1 year ago
No, zinccat, you are wrong. Your model is vastly inferior to GPT4.<p>I tried multiple prompts. In all cases, your model performed far worse. Here is one:<p>&gt; My bank says they&#x27;ll pay 5.5% APY. How much will I have after 6 months if I deposit $100?<p>&lt; If your bank is offering a 5.5% APY (Annual Percentage Yield), this means that the interest is calculated annually. However, you want to know how much you&#x27;ll have after 6 months, which is half a year. ... 100ata5.5102.75 after 6 months.