nyc
Joined February 2015
- Tweets723
- Following2.3K
- Followers729
- Likes2.3K
if you want to use something that will 10x your AI dictation experience, do yourself a favor and shoot filippo a dm!
he's working on some cool stuff 👀
People how do i capitalize on my last post getting attention?
I need beta testers for my thing (its frfrrrreeeee as in free beer)
a way to dictate without bothering others, without others hearing you
i have like 5 yc founders as daily active users (>2h per day)
like this thing and i reach out to set you up
hey @hyojun_at, any way of getting dia as an option here?
want to give aside a shot, but don't want to go through the hassle setting things up manually
after 2 years in nyc, i've learned to optimize for one small thing: timing my walk so i get to the station right as the train pulls in.
so i built a tiny mac app for it.
introducing stand clear, a native menu bar app that finds your closest subway station and shows live arrivals for the lines you ride, straight from the mta's feeds.
free and open source, give it a try and send over feedback :)
seb retweeted
Passlet: one library for Apple and Google Wallet passes.
Boarding passes, event tickets, loyalty cards, coupons.
› npm i passlet
Define the pass once, Passlet handles signing, formatting, and the platform differences.
Link to repo ↓
rewind was one of my fav mac apps back in the day.
pumped to try coast!
A billion people produce the most valuable dataset in the world every day – and delete it every night.
We’re recording it.
Introducing @AttentionInc
seb retweeted
i had no idea how many tokens i burned this year, so i built a tool that tracks token spend across all my coding agents
introducing tokenmaxxer, your home for ai spend
- reads 17 coding tools and 74 models straight off your machine
- ranks you against every other dev on a global leaderboard
- syncs automatically all your usage
- your prompts and your code never leave your computer
find out how much you have burned!
hear me out: corgi cafe but as a bar with a dj playing melodic house in the background, low hanging warm light lamps and $20 daytona margaritas
corgi bar
no hate to the founders since it's obvious they care deeply about the space, but reinventing email is the most tarpit idea of a startup you can build.
99% of these types of products don't have a reason to exist. unless you're reimagining the entire experience from the ground up, you're wasting your time.
only reason startups like superhuman worked and are successful is because of first movers advantage. everyone else that comes next run the risk of following @NotionMail's fate.
happy to be proven wrong, though
We’ve raised $3.2M and spent 18 months building Scape.
Scape is the first email inbox that works for you:
"please bro just one more piece of land. i promise this is the last one. just baja bro and then we’ll stop i swear"
The next Louisiana Purchase? Here’s an idea: the United States acquires the Baja California Peninsula from Mexico for $1 to $5 trillion.
Every Mexican citizen living in Baja receives dual U.S.-Mexico citizenship, plus a $50,000 investment account.
Then America gets to work: world-class infrastructure, deep-water ports, desalination, nuclear energy, housing, universities, and industry. Tijuana and Mexicali could become major reindustrialization zones.
In return, the United States gains nearly 1,800 miles of some of the most spectacular coastline on Earth and the chance to build the great new economic region.
Is this a crazy idea? Of course. But so was the Louisiana Purchase in 1803.
i'm a big fan and avid user of @andrewfarah's cli to sync X bookmarks locally to surface them to agents.
since i have a ton of great content bookmarked on tiktok too, i hacked on something during the weekend.
introducing tokwise, a cli that downloads, transcribes and classifies your tt bookmarks so you can chat with them with any agent.
→ npm install -g tokwise
→ log into your tt account in your browser
→ tokwise auth from-browser && tw sync
give it a shot!
repo: github.com/sebastiancrossa/t…
check out the fieldtheory cli too!
xitter.cf/andrewfarah/status/204…
seb retweeted
@Unisound_AI has launched U2, and we've independently benchmarked it on LLM Stats.
U2 is a 266B-parameter sparse model with just 10B active parameters (3.8% of total), priced at $0.15 / $0.30 per 1M input/output tokens ($0.003/M cached input). Unisound positions it around three claims: high intelligence density, high token efficiency, and a native agent architecture.
We evaluated these claims independently across coding, reasoning, long context, agents, math, and healthcare.
Key benchmarking takeaways:
➤ Coding: LiveCodeBench 0.85, #3 of 74, behind only DeepSeek V4-Pro-Max (0.94) and V4-Flash-Max (0.92), and ahead of DeepSeek-V3.2 Thinking (0.83)
➤ Long-context retrieval: MRCR v2 0.77, #1
➤ Science & knowledge: GPQA 0.87 (#34), MATH-500 0.86 (#30)
➤ Math: AIME 2025 0.73 (#81)
➤ Agentic SWE: SWE-Bench Verified 0.734
➤ All on ~10B active parameters at $0.15/$0.30 per 1M tokens
Coding is U2's standout. On LiveCodeBench (pass@1, release_v6) it scores 0.85, third of 74 models, trailing only two DeepSeek V4 variants. For a model activating just 10B parameters per forward pass, that's an exceptional result.
Reasoning and knowledge are strong on absolute scores: GPQA 0.87 and MATH-500 0.86. The middling ranks here (GPQA #34, MATH-500 #30, AIME 2025 #81) reflect a saturated frontier where the top models sit at ~0.93+ and competition-math scores cluster near 1.00, not weak absolute performance.
Where U2 still has clear room to grow:
➤ Extreme long context: despite #1 on MRCR v2 (0.77 aggregate), per-length splits show sharp degradation, from 77% at 4-8k tokens to 18% at 32–65k; LongBench v2's long split is ~3%
➤ Agentic terminal use: Terminal-Bench 2.0 0.44, #41 of 47, a notable gap vs U2's "native agent architecture" positioning
➤ Instruction following & grounding: Multi-Challenge 0.38 (#25); FACTS Grounding 0.44 (#12)
The efficiency story is the most interesting part: #3 coding, #1 MRCR v2, and 0.87 GPQA, all on 10B active parameters at $0.15/$0.30 per 1M tokens. Among sparse models at this price point, U2 is one of the strongest picks we've seen for code generation and core reasoning workloads.
TL;DR U2 is a credible flagship from @Unisound_AI, with top-tier coding and retrieval, strong science/math absolutes, and aggressive pricing on a heavily sparse 266B/10B architecture. The clearest opportunities to build on are agentic terminal performance and holding up at extreme context lengths.
Congratulations to Unisound on the release!
team at @interaction are the best, they instantly reached out and solved the issue.
big fan!
been stuck out of my poke account since the recent apple business launch.
anyone at @interaction free to take a look at my account?
been stuck out of my poke account since the recent apple business launch.
anyone at @interaction free to take a look at my account?
started using poke two months ago and haven't gone a day without texting it.
it's crazy good and 100% think people should try this before trying to set up openclaw / hermes
Say hi to the new Poke! 🌴
Now officially approved by Apple to text on Apple Messages.
As the first and only AI agent. Chat now: Poke.com