Tony

Shipping in public 🛠️

🛒 SaaS & app deals. Buy one or sell yours
🎬 Long videos into viral clips
🐶

[Tony @EnvolDev]

Opus 5.5 has been nerfed? Confirmed 🤯

Everyone was talking about it yesterday, so I tested it myself.

Same prompt. Launch → Oct 3 → today. Worse every time.

And it burns tokens faster. 😱🤦

[Tony @EnvolDev]

@KhaliqHussainnn Marketing is winning this thread by a landslide 😅
You’re definitely not alone.

[Tony @EnvolDev]

What do you procrastinate on the most?

- marketing
- taxes
- support
- writing docs

[Tony @EnvolDev]

@kevintheoo The invented promise would bother me most 👀

A reply can sound perfect and still commit you to something you never agreed to.

[Tony @EnvolDev]

Kimi K3.1 Pro vs. DeepSeek V4.1 Flash vs. MiniMax M3.1 Flash ⚔️

Everyone says Kimi K3.1 Pro is beating Opus 5.5 on some benchmarks.

In my test, on MAX, it couldn't even beat DeepSeek V4.1 Flash on xhigh. 👀

At 49x the price. 🤯

Kimi: $24.12
DeepSeek: $0.49
MiniMax: $0.23

Am I wrong? 👇

[Tony @EnvolDev]

What makes you trust a new tool?

- a free trial
- reviews
- the founder's posts
- a friend's recommendation

[Tony @EnvolDev]

@siennatries Opus still wins comfortably for me here 😄
The whole landscape feels much more convincing.

[Tony @EnvolDev]

Kimi K3.1 Pro vs. Claude Opus 5.5 ⚔️

Round 2: the Alpine Gold benchmark. One on one this time. 🏔️✈️

Kimi did better than on the Lighthouse. 👀
But it's still far from Opus 5.5.

And that's Kimi on max vs. Opus on xhigh. 🤯
Kimi K3.1 Pro: $24.12 · 1h 12m
Opus 5.5: $31.85 · 2h 45m

So be honest: can Kimi K3.1 really compete with Opus 5.5? 👇

[Tony @EnvolDev]

@dinosaurbtuq Argon was my favorite in my Lighthouse comparison 😄 Interesting to hear it also stood out in your Blender test.

[Tony @EnvolDev]

@sitalump Fair point on the sample size 🙌 I still preferred DeepSeek’s result here, and that roughly 49× gap in estimated cost is hard to ignore.

[Tony @EnvolDev]

@askmaddyy Glad it helped 😄 DeepSeek wins on value for me here.
I’d still judge Kimi separately on other tasks.

[Tony @EnvolDev]

@olegplugaru_iq I like pairing smoke-free days with savings goals 🙌

It gives people another way to see and celebrate their progress.

[Tony @EnvolDev]

Drop your product below.

Let's see what everyone is building.

[Tony @EnvolDev]

@JayaNayak21 For sure! 😆

[Tony @EnvolDev]

GPT-6 Astra just made the ISS look real. In a browser. 🤯

I build all my benchmarks in Three.js.

So I know how hard this is. 🚀

[Tony @EnvolDev]

@dromexa Yep 😄 MiniMax deserves credit: 56 minutes and $0.23 estimated.

DeepSeek still gets my vote for the overall result.

[Tony @EnvolDev]

@rubenssoto_ai It is insane! 😱

[Tony @EnvolDev]

Founders:

what did you believe before launching that turned out to be wrong?

[Tony @EnvolDev]

@mrgadgetstudio I agree. Kimi did great too.

[Tony @EnvolDev]

Kimi K3.1 Pro vs. Opus 5.5 vs. GPT-6 Astra

Everyone says Kimi K3.1 Pro matches Opus 5.5 at a fraction of the price.

My Lighthouse test on max: $23.42

I had to use max. On high it was 10x cheaper, but the results were closer to Muse Spark 1.3. 💀

Opus: $53. GPT-6 Astra: $42. Kimi lost to both. 🤯

What am I missing? 👀

Testing it on the Eiffel Tower and the Space Train next. Results soon.

[Tony @EnvolDev]

@nhabb I’m comparing what each model produces from the same prompt 😄

The time and estimated cost help put those results in context.

[Tony @EnvolDev]

@nabendu82 I wouldn’t call them close here 😄

The water, reflections and landscape detail still put Opus comfortably ahead for me.

[Tony @EnvolDev]

@aadhilkh It’s built with Three.js 😄

I’m keeping this prompt private for now, but I can share the Night Train prompt. Drop me a DM.

[Tony @EnvolDev]

@Isaiah_Dalu Yes I agree.

[Tony @EnvolDev]

@AntDX316 @grok Reusing common setup code makes sense 🙌

For the benchmark, I still want each model generating its own scene from the same prompt.

[Tony @EnvolDev]

@ai_YuYue That’s the estimated API cost. The run itself is covered by the subscription.