GPT-5.6 Sol vs Fable 5 both models were tested on the same reasoning with the same prompt > 5.6 Sol one shotted everything in 70 minutes in devin > Fable 5 one-shotted everything in 90 minutes in claude code really surprised by Sol results here which one do you... think won?show more

front test is where fable 5 is really good, decided for yourself both were given same prompt Fable 5 Max Sol 5 Max

on how many tokens each used > fable max used 25% of session limits and 10% weekly limits for both front end and minecraft on 200$ max plan > devin don't have session limits so it used like 20% of weekly limits for both front end and minecraft on 200$ max plan verdict: sol is fast but uses more tokens and give way better results fable is slow but uses less tokens and give good results decide for yourself

working shadows, wow some minor texture issues on the hand, but other than that i liked sol more

haha yess, villages and alot more was added by sol

what!! 😳 has day/night and even rainfall an inventory too sol definitely won this

yeah, even added mobs and caves too I'll push game on vercel and share link so you guys can test

that would be sickkk

Sol did well but both were amazing

Crazy one man ..

Which one do you like?

Favor had very clear blocks and movement but both are detailed as f.

Ah kk

Gpt more better

Lives up to hype

Chatgpt did a better job overall!

For frontend use too it so good and best

Sol cooked

Gpt 5.6 Sol for me

same, it cooked harder

The only SOL that is working right now

bro is the biggest open ai ambassador haha

hey @thsottiaux can I be ambassador for OpenAI from pak??

@thsottiaux whoever you are, believe me, you won't find a bigger fan than him

i like both

Yo love to see this, means we are getting more and more great models so we would greatly benefit from the competition!

this is close asf

they look well matched but Fable seems to nudge it with more attention to the finer details🧐 but with that said, with some iteration im sure Sol could easily meet the same mark

@LLON3RR Sol defitnely cooked here damnnnn gpt takeover

the honest answer is that these models are getting closer to a qualitative advantage, but it's still unclear how they'll perform on edge cases and complex real-world scenarios until we test them there.

SOL definitely won this round but I think fable might just thrive with a different prompt or request.

OpenAI cooking again

they deserve it sol lives up to hype

not bad but fable 5 result looks more better for me

Sol did good

@Brookcalls really good but not same as fable

where is tha who was building gta 6

Fable literally copied Minecraft's look. Sol, come up with something of his own.

something different and unique

5.6 sol cooked ngl

What model you using mostly nowadays?

Fable and soln

Honestly, I think we’ve reached a point where models already work well. With any of the newer ones, if you keep iterating, you can get almost anything done pretty easily.

The real variable here is environment scaffolding, not model reasoning. Devin's tool loop bails out faster on ambiguity; Claude Code churns until it hits a wall. The win is whoever fails less silently.

really incredible for both

got cooked it hard

5.6 Sol mogged Fable 5 easily

I like how fable 5 was done better but the speed of chatgpt 5.6 is amazing

oh wow the improvement is crazy

70 minutes? CRAZY!! 5.6 def won!