Arnav Gupta @_ar9av

Tweets

725
Followers

179
Following

613
Likes

518

Arnav Gupta @_ar9av

a day ago

I think the best way of using Claude 4.5 or GPT Codex is in a Docker container! Isolate your env, mount your project and skip perms safely

1 0 2 58 0

Arnav Gupta @_ar9av

a week ago

With OpenAI getting $100B from Nvidia, Anthropic must move fast to secure similar funding or risk falling behind on compute. Musk is driving Colossus 1 & 2 Google sits on cash + TPUs The race is on

1 0 2 62 0

hotlapdaily @hotlapdaily

3 weeks ago

Driven by community! Thanks for all the feedback

0 1 3 218 0

Download Image

Arnav Gupta @_ar9av

a month ago

UI suggestions to improve?

0 0 8 160 2

Download Image

Arnav Gupta @_ar9av

a month ago

"we gotta listen to our users and deliver!" Our MAU was down, and since then @hotlapdaily grew by +50% in just 30 days Here's the trajectory! 20k MAU (launch month) 15k MAU (2nd month) 30k MAU (3rd month) Since then our followers on X +110% Discord community +200% Happy to…

0 0 3 144 1

Download Image

Arnav Gupta @_ar9av

a month ago

LLM evaluation is genuinely a big problem Using one LLM to validate another sounds smart but it’s tricky: cascading errors, conflicting biases, compute costs, latency There is no guarantee the validator catches subtle hallucinations I wonder how to solve it

1 0 2 92 0

Anish @anishmaybe

2 months ago

Things are getting out of hands for @hotlapdaily We don't know who these users are but one thing is 4 sure!

0 1 7 172 0

Download Image

Arnav Gupta @_ar9av

2 months ago

I have been having fun reading these trychai.io/blogs/gists

1 0 2 84 0

Arnav Gupta @_ar9av

2 months ago

Built @hotlapdaily Its now having 1000+ DAU :) No login/ads/cookies game. Will follow anyone who beats my time Happy to share my GTM strategy

2 0 7 165 0

Download Image

Arnav Gupta @_ar9av

2 months ago

Everyone is hyping OpenAI’s open models, but for me my tests (on tool calling and 0-1 app dev) show Qwen-3-32B and GLM-4.5-Air still edge out in speed and robustness for coding and tool use SOTA isn’t a single leaderboard, test on your own workflows before making the call!

0 0 1 134 0

Arnav Gupta @_ar9av

2 months ago

After 50 days of using Meta Glasses

1 0 4 666 0

Download Image

Arnav Gupta @_ar9av

2 months ago

what is the first productivity app idea that comes to your mind?

3 0 10 560 0

Arnav Gupta @_ar9av

2 months ago

There is something wrong with Docker desktop MCP function with Claude It keeps on disconnecting (but sometimes it works) @Docker @AnthropicAI

0 0 2 150 0

Download Image

Arnav Gupta @_ar9av

2 months ago

100M context window 😮‍💨 But past 10% it breaks lmao same playbook as 1M - carrot in front, stick behind They be selling dreams till the next shiny number drops lol ok

0 0 4 120 0

Download Image

Arnav Gupta @_ar9av

2 months ago

If you're running diff SLMs for fine-tuning + hosting, always check MIG (Multi Instance GPU) support before buying your next GPU MIG lets you slice GPUs into isolated instances (small mini GPUs) perfect for parallelism + resource control. Not all GPUs support it. Check specs.