Is it just me or do you guys also feel like nowadays your entire worth as a developer is at the will of OpenAI and Anthropic?

@andrhlt

10.3K followers · usually 5.9K views a post

Views

124.7K

Likes

490

Comments

45

Shares

10

Selling · SaaS › AI Tool

Kimi K2.5seen in 4 videos →https://kimi.moonshot.cn/ ↗

Why it worked

The video taps into a genuine developer pain point (vendor lock-in and rising costs) and provides a specific, actionable alternative, making the product mention feel like a helpful recommendation rather than an ad.

Summary

The creator discusses the risks of relying on closed-source AI models for coding and advocates for open-weight alternatives. He highlights Kimi K2.5 as a high-performing, cost-effective open-weight model he recently used for development tasks.

Structure

1Identifies the problem of dependency on big AI providers
2Explains the risk of price hikes and efficiency loss
3Introduces open-weight models as the solution
4Shares personal experience using Kimi K2.5
5Compares Kimi K2.5 performance and cost to Opus
6Concludes with a bullish outlook on open-weight AI

Original caption

Why I’m coding with open weight models #cursor #claude #kimi #coding #developer

More from @andrhlt

Their other posts in the index, biggest breakout first.

Product placement

The creator explicitly mentions and praises Kimi K2.5 as a superior, cost-effective alternative to closed-source models like Claude/Opus.

On-screen text

Why I'm coding with open weight models

Transcript

Is it just me or do you guys also feel like nowadays your entire worth as a developer is at the will of OpenAI and Anthropic? I rely on AI so much for coding now that what happens if OpenAI or Anthropic is to like 5x their prices? And what if all the other developers are totally cool paying that 5x price and you don't want to? Are you just like 20% less efficient than everyone else? Sounds crazy, but they have the control to do this. And this is why I'm such a huge fan of the open weight models. Which was pretty discouraging until recently cause we just had Qwen 3 and GLM 4.7 which I would try coding with them and they would honestly kind of just tear my codebase to shreds. The first open weights model I've ever used and has actually been competitive with the closed weight models is Kimi K2.5. I was doing a lot of agentic coding with Kimi K2.5 this weekend and it was doing difficult tasks for like 10% the cost of what it would have taken Opus. It can definitely hold its own. And it's so incredible how much intelligence we can now have for so cheap. This is why I'm super bullish on open weight models. Why I feel like I have to be super bullish on open weight models. Super excited to see what labs like Kimi do in the future. And it's showing that the closed weight models aren't necessarily going to be the highest performing ones.