Vinny Barreca (@PhantomByteAI) header
Vinny Barreca (@PhantomByteAI) profile picture

AI tools, bots & lead-gen machines for solo operators and small teams. I publish what works, then automate the rest. Code from the shadows 👻

Joined May 2015
Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)Photo by Vinny Barreca (@PhantomByteAI)
80% of your agent's tokens are heat. arXiv 2611.23375: 2,305 tokens per change. 1,890 heat. 351 free energy. Fuel, not distance. Which call moved the diff? deadweight: 61.2k tokens before you type. articles.phantom-byte.com/co…
1
1
103
The biggest IPO ever sold off. Anthropic opened $111.50, closed $97. $81.3B raised, $1.3T valuation. Revenue $16.2B in H1 2026 vs $4.6B in 2025. Taiwan -6.7%. The market priced it, then said no. What is $1.3T underwriting? AI hardware is now a tenth of world trade. Rates rose.
1
1
116
Replying to @thinkymachines
You guys nailed it with Inkling 2 small and 2 mini. 🔥
43
Replying to @AlexFinn
Small and small mini, they are two models, both awesome.
422
Replying to @ashxhart
Inkling 2 small is a great model, so there is no doubt in my mind this is a great little model.
1
43
Replying to @skirano
Inkling 2 Small is a great model also. I've been using it since it's been out.
1
310
Replying to @yeahfortommy
I've been using Hermes Agent since the beginning. I use mostly Deepseek, BUT Inkling 2 Small is my second most used model.
1
78
Flash is dead. Good. Google retired 3.8 Flash for Gemini 4 Hydrogen: Terminal-Bench 4.1 up 18.2% to 57%, 2.5x faster, $1.50/$7.50 per million over a 1M window. Cheap and fast is what ships. What are you rebuilding? Frontier wins headlines. Fast tiers win users.
78
272 fine-tuning runs, 7,382 GPU-hours measured. The standard FLOP estimate explained 31% of real wall-clock time. Mean error: 12.41 hours on 23.43 hour jobs. The math isn't wrong. The model is. dev.to/phantom-byte/why-your…
1
77
Replying to @RandallKanna
That’s false. I’ve been coding since 1993, and I love my 7+ agents.
13
Replying to @miramurati
Well done 👏
1,095
Replying to @boardyai
Love it 😍
1
6
Replying to @boardyai
If I start building in public, a bunch of people are going to get a vocabulary lesson. When I’m done, you’ll know every four-letter word ever known.
1
27
Replying to @realtalkstruth
Mine are mostly Hermes agents, there are 7 of them, one Meta Muse, a Codex agent, 2 pi coding agents, and one OpenClaw.
1
68
Replying to @Manhattva
Right? Anyone who has managed people has wanted to say this at some point. People forget, Jensen also refused to fire people, his words not mine; he’d rather torture them to greatness.
2
44
Replying to @EricLDaugh
The issue is the American labs, with the exception of a few open-source American labs, can’t be trusted. And in the case of the OpenAI hacks, they needed these Chinese models to fight them off because American labs refused to help.
299
Replying to @DailyMail
Daily Mail has always been junk.
4
149
Stop renting a cloud to test. AWS open-sourced an emulator for 200+ service APIs, local HTTPS, no cloud spend. Set AWS_ENDPOINT_URL, leave your code untouched, CC0 on GitHub. Every agent team needed this. What still blocks your local loop? Fast feedback beats a big bill.
1
125
180B parameters. Laptop RAM. Hugging Face merged expert streaming into Transformers: weights stream off SSD, within 6% of Strata. Qwen3.8-Flash-Next runs on a 32GB Mac. Weights stopped being the moat. What are you running at home? Cloud rent ends. Local leverage begins.
1
1
161