Pankaj Kumar (@pankajkumar_dev) - Profil Twitter

Tweet Disematkan

Pankaj Kumar@pankajkumar_dev·27 Nis

Projects that I have made in last 6 months! feedwall.vercel.app flowpay-one.vercel.app boltweb-ai.vercel.app ui-unify.vercel.app resume-sach.vercel.app trimmrr.vercel.app vimal-parody.vercel.app Learned a lot while making these projects ,huge thanks to @kirat_tw for the amazing teaching! 🙌

English

148

95

2.2K

242.3K

Pankaj Kumar@pankajkumar_dev·12h

@Mk__0168 idk , their internal codename

English

0

6

Mahendra Kumawat@Mk__0168·12h

@pankajkumar_dev What's the story behind these codenames? Are they related to specific features or projects within Google Gemini? Curious to know more about the tech behind the names

English

1

0

2

7

Pankaj Kumar@pankajkumar_dev·15h

Google internal naming in Antigravity Model Names >> Actual Power Levels

English

3

2

9

731

Pankaj Kumar@pankajkumar_dev·12h

@devsharmatwt idk 🤣

0

1

10

Dev Sharma@devsharmatwt·13h

@pankajkumar_dev What the hell is INFINITY_JET 🤣

English

1

0

1

31

Pankaj Kumar@pankajkumar_dev·15h

@anjany06 lets hope.

English

0

41

Anjaney@anjany06·15h

@pankajkumar_dev Banger

Indonesia

1

0

1

68

Pankaj Kumar@pankajkumar_dev·15h

x.com/pankajkumar_de…

Pankaj Kumar@pankajkumar_dev

Google internal naming in Antigravity Model Names >> Actual Power Levels

ZXX

0

1

97

Pankaj Kumar@pankajkumar_dev·2d

Antigravity, what are you guys doing? - I was waiting for today because the Claude model quota was supposed to reset so I could use it again. - But now its been extended by another 6 days… again. - I haven't been able to use Claude models for the past 2 weeks. At this point, it feels like we can't use them at all. - If this continues, its better to just remove Claude models entirely and now give Pro users an option for a refund on the remaining subscription. Right now, it just feels like a wasted subscription for those who paid specifically for Claude access.

English

304

87

1.3K

101K

Pankaj Kumar@pankajkumar_dev·15h

@anjany06 Ok

0

21

Anjaney@anjany06·16h

Hey guys, busy with internals right now. What are you all building? If you're into tech, let’s hit 500 verified followers before March ends

English

38

0

35

948

Pankaj Kumar@pankajkumar_dev·1d

ZXX

0

185

Pankaj Kumar@pankajkumar_dev·1d

GPT-5.4 Mini & Nano: The Cure for Burned Quotas and High Costs - The Quota Problem: GPT-5.4 is insanely powerful, but running it on long tasks or agents quickly drains your weekly limits and racks up API costs. - The Mini Advantage: GPT-5.4 Mini cuts usage to ~30% of the flagship, making heavy workflows around 70% cheaper without changing how you build. - Near-Flagship Performance: The trade-off is minimal Mini scores 54.4% on SWE-Bench Pro vs 57.7%, and 72.1% on OSWorld vs 75.0%. - Nano = Scale Mode: GPT-5.4 Nano is built for volume, priced at just $0.20 per 1M input and $1.25 per 1M output tokens. - Cheap but Capable: Even at that price, Nano delivers 52.4% on SWE-Bench Pro and outperforms Claude Haiku 4.5 (41.0%). - The New AI Stack: Instead of one expensive model, teams are orchestrating GPT-5.4 for planning, Mini for execution, Nano for fast background tasks. Availability: Mini is live across ChatGPT, Codex, and API, Nano is currently API-only.

English

7

1

27

3.9K

Pankaj Kumar@pankajkumar_dev·1d

@shekhar_tw sad, so u use claude code than ? its the best but pro plan has very low limits.

English

1

0

1

24

shekhar@shekhar_tw·1d

@pankajkumar_dev Ohh, its not working for me🙂

English

1

0

1

15

Pankaj Kumar@pankajkumar_dev·1d

@shekhar_tw its good better than antigravity and the ratelimits is good even in go plan.

English

1

0

1

55

shekhar@shekhar_tw·1d

@pankajkumar_dev review?

English

1

0

1

15

Pankaj Kumar@pankajkumar_dev·1d

@shekhar_tw yess

English

1

0

1

70

shekhar@shekhar_tw·1d

@pankajkumar_dev are u using codex?

English

1

0

1

82

Pankaj Kumar@pankajkumar_dev·1d

@CChirchi hmm, lets see how they perform.

English

0

48

Chahid Chirchi@CChirchi·1d

@pankajkumar_dev finally quotas won't kill my budget!! thx for sharing

English

1

0

1

55

Pankaj Kumar@pankajkumar_dev·1d

@pxrtal i think, it would be we have to test it but as i am seeing the flagship models are too expensive , so iam gonna support these cost effective models more.

English

0

101

Luis Portal@pxrtal·1d

@pankajkumar_dev do you feel that it is really optimized for code?

English

1

0

1

131

Pankaj Kumar@pankajkumar_dev·1d

@10xshivam yeap

English

0

1

100

shivam@10xshivam·1d

@pankajkumar_dev soon across all platforms

English

1

0

2

131

Pankaj Kumar@pankajkumar_dev·1d

@10xshivam enjoying.

English

1

0

1

34

shivam@10xshivam·1d

what do you feel when you don’t open your code editor for a whole day because of exams

English

1

0

7

159

Pankaj Kumar@pankajkumar_dev·1d

x.com/pankajkumar_de…

Pankaj Kumar@pankajkumar_dev

GPT-5.4 Mini & Nano: The Cure for Burned Quotas and High Costs - The Quota Problem: GPT-5.4 is insanely powerful, but running it on long tasks or agents quickly drains your weekly limits and racks up API costs. - The Mini Advantage: GPT-5.4 Mini cuts usage to ~30% of the flagship, making heavy workflows around 70% cheaper without changing how you build. - Near-Flagship Performance: The trade-off is minimal Mini scores 54.4% on SWE-Bench Pro vs 57.7%, and 72.1% on OSWorld vs 75.0%. - Nano = Scale Mode: GPT-5.4 Nano is built for volume, priced at just $0.20 per 1M input and $1.25 per 1M output tokens. - Cheap but Capable: Even at that price, Nano delivers 52.4% on SWE-Bench Pro and outperforms Claude Haiku 4.5 (41.0%). - The New AI Stack: Instead of one expensive model, teams are orchestrating GPT-5.4 for planning, Mini for execution, Nano for fast background tasks. Availability: Mini is live across ChatGPT, Codex, and API, Nano is currently API-only.

ZXX

0

531

Pankaj Kumar@pankajkumar_dev·2d

@amanpandey7681 @charliextwo it will end after 5-6 prompts

English

1

0

1

77