GingaNinja

[GingaNinja @HumanMeatloaf98]

@Tech_girl Then all corporations would need to find a new place to store their code lol.

[GingaNinja @HumanMeatloaf98]

@thsottiaux @maria_rcks Please don’t remove this feature. Don’t let bad people take this from me. It’s so much nicer than Claude

[OpenCode @opencode]

Ox Alpha (stealth model) is free for the next week

- 1M Context
- Multi-modal
- Zero Data Retention

Generous rate limits, near unlimited usage

We have capacity for 100T tokens per day, lets see what you can do

[GingaNinja @HumanMeatloaf98]

@dreamsofcode_io Ahh to learn then yeah I’d agree with your OP. But there comes a “switch” where manual coding would be a “waste of time”. Like for me, I’d consider manually coding a waste of my and my companies time.

[GingaNinja @HumanMeatloaf98]

@dreamsofcode_io I mean I’m biased as I have 6yrs swe experience before ai. So I’d take someone who had that experience but amplified with ai. I wouldn’t want a swe manually coding for me.

[GingaNinja @HumanMeatloaf98]

@paul_e_jones @CityCoderUS lol that’s great.

[GingaNinja @HumanMeatloaf98]

@bridgemindai Wild to me Luna costs Pennie’s compared to opus 5 yet is so much better at swift at least.

[GingaNinja @HumanMeatloaf98]

@bridgemindai I legit had Luna fix opus 5 mistakes. Opus 5 put a test case in that actually solidified a bug. I had Luna review opus 5 code and gave opus 3 attempts to fix based off Lunas review. Had to just have Luna do the fixes. Had opus learn from it.

[GingaNinja @HumanMeatloaf98]

Had GPT Luna fix opus 5 code. Then had opus learn from it. Opus 5 is trash.

[GingaNinja @HumanMeatloaf98]

@nickmmark @kimmonismus @gorillapanics It is a supervised machine learning algorithm utilizing deep neural networks. It’s not a LLM tho, that’s correct. It was locked in pre transformer based models (2018/2019) for fda approval to move into clinical trials.

[GingaNinja @HumanMeatloaf98]

@andrewaiginin @klein_activist Gemini - Neither architecture unilaterally dominates biomolecular design; rather, diffusion models dominate 3D spatial backbone generation, while transformers and Protein Language Models (pLMs) dominate 1D sequence design, property optimization, and functional modeling.

[GingaNinja @HumanMeatloaf98]

@bridgemindai My opinion, they all suck lol. I still have to guide them so much to get good UIUX. And some UIUX I actually yell at them for, like “what made you think this was a good idea?” lol.

[GingaNinja @HumanMeatloaf98]

@wholemars We just need to innovate on the whole process it takes from discovery to fda approval. That is what takes decades. I’m hopeful we can have successful human body simulations in the future where we can run initial clinical trials in at rapid speed.

[GingaNinja @HumanMeatloaf98]

@klein_activist @andrewaiginin New specialized models are LLMs tho. Lots are now transformer-based architectures.

[GingaNinja @HumanMeatloaf98]

@Umesh__digital People act like ai writes the code but doesn’t debug. Debugging is as simple as another prompt lol.

[GingaNinja @HumanMeatloaf98]

@neerajjj6785 Disagree. Define “good developer”. I was a senior swe before AI and now write 0 code. Am I worse at coding? Absolutely. Am I a worse engineer? I’d say im a better engineer than before. I just worry about other things than nitty gritty code details.

[GingaNinja @HumanMeatloaf98]

@sl1ma4 Yes.

[GingaNinja @HumanMeatloaf98]

@solopribuilds Ehhh, there’s nothing unique in my data that they don’t already have / know guaranteed.