lyra

hihi haha @

[lyra @lyraxana]

[GPT Astra] BMW M4 Competition SVG
~25k tokens; ~19 minutes; effort: max
[Gemini 3.1 DeepThink] BMW M4 Competition SVG
~3.3kk tokens (~458k reasoning); ~29 minutes; thinkinglevel: high
[Gemini 3.8 Flash] BMW M4 Competition SVG
~19k tokens; ~42 seconds; thinkinglevel: high

[lyra @lyraxana]

@XIVIX_134 @thsottiaux Astra

[lyra @lyraxana]

@theowensong no they weren‘t - from me

[lyra @lyraxana]

@Ceoz_1 @chetaslua it does

[lyra @lyraxana]

[GPT Astra] Peacock SVG
~38k tokens; ~14 minutes; effort: max

[lyra @lyraxana]

@balubabear1312 it's not about showing the "best" possible outputs on my page, but yes I agree it's not looking good but the physics are great.

[lyra @lyraxana]

[GPT Astra] CS Surfing
~28k tokens; ~25 minutes; effort: max

[lyra @lyraxana]

@kimmonismus @IliaBudylin it is fake

[lyra @lyraxana]

@YwspRy38087 then I will never see an output

[lyra @lyraxana]

[GPT Astra] Death Star
~75k tokens; ~26 minutes; effort: max

[lyra @lyraxana]

@YwspRy38087 zero shot

[lyra @lyraxana]

@DavidSZD1 yes 2K and 4K are accepted on instant-ramen-v2, but not on instant-ramen (current Nano Banana 2 Lite)

[lyra @lyraxana]

1. [Nano Banana 2 Lite] (instant-ramen)
~2.5k tokens (~800 reasoning); ~8 seconds; thinkinglevel: high
2. [Nano Banana 2 Lite UPDATE] (instant-ramen-v2)
~3.1k tokens (~1.6k reasoning); ~19 seconds; thinkinglevel: high

[lyra @lyraxana]

[GPT Astra] Ana de Armas TiKZ (LaTeX)
~42k tokens; ~15 minutes; effort: max
[GPT Astra] Jessica Alba TiKZ (LaTeX)
~43k tokens; ~16 minutes; effort: max

[lyra @lyraxana]

@TimJayas agreed

[lyra @lyraxana]

[GPT Astra] Pelican Voxel
~89k tokens; ~30 minutes; effort: max
[GPT Astra] Pelican SVG
~31k tokens; ~23 minutes; effort: max

[lyra @lyraxana]

@Deepusleepy hi yes

[lyra @lyraxana]

[GPT Astra] Pagoda Voxel
~56k tokens; ~38 minutes

[lyra @lyraxana]

@Szpadel0 @chetaslua exactly

[lyra @lyraxana]

@kecksbe forgot to mention the reasoning effort: Max