i build with the models everyone’s arguing about

Joined September 2014
tripcop retweeted
We’re committing a portion of Anthropic’s shares to the Anthropic Public Benefit Fund, a new wholly owned entity dedicated entirely to the public benefit. It will start off with 5% of our shares. For every 50% increase in the company’s value over a set baseline, we can contribute up to an extra 2%, subject to our board’s discretion. In total, this means that roughly 20% of Anthropic’s shares are committed to the public good. Our co-founders have already pledged to donate 80% of their equity, and around half our employees are donating a portion of their equity as part of a donation matching program we’ve run since our founding. In the post below, we share more details about our new fund. anthropic.com/news/public-be…
2,177
1,652
1,589
20,722
6,953,468
tripcop retweeted
Vote for Soulja Boy 2028 🇺🇸
566
1,072
294
7,177
274,276
tripcop retweeted
Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released. On average, it costs around 75% less to run than Claude Haiku 4.5.
1,609
3,275
2,427
45,255
6,679,461
Wtf
government of India officially removes bitchat from the App Store
33
Buzz is the most interesting agent infrastructure shipping right now. humans plus agents, building together, on infra you own. and the AI community is completely silent about it
🐝 Buzz Desktop v0.5.26 is here! This release brings more reliable conversations, better sidebar syncing, and new tools for relay operators. Your sidebar stays in sync. Fixed an issue where sections, sorting, starred items, and mutes could show outdated settings when opening Buzz on another device. Better handling of long conversations. Improved loading of replies and reactions in long threads, with better timeout handling when a query takes too long. A new relay admin console. Authorized operators and moderators can review reports and product feedback directly in Settings. Operators can also manage the operator and moderator roster. Smoother project approvals. Fixed channel-creation approvals for unlisted project home channels. A corrected setup link. The Grok Build setup guide now points to the right place. Under the hood, this release also adds relay support for channel artifacts and native encryption support for private-key backups. github.com/block/buzz/releas…
1
1
1
210
Nobody believes this marketing bullshit anymore.
BREAKING: OpenAI scraps GPT-6.1 Astra release over safety concerns
43
Crazy idea, huh? @amazon @Walmart and the rest of the big dogs
BREAKING: Costco is using tariff refunds to cut prices on everyday goods
93
The National Weather Service needs to put some fucking pants on
A MODERATE risk is in effect in our Day 3 Excessive Rainfall Outlook. More details: wpc.ncep.noaa.gov/qpf/excess…
1
124
Fuck yeah m
20
UFC became WWE at this point
Luis Hernandez celebrates his first round submission victory of Sedriques Dumas with a donut 👮🏻‍♂️🍩
1
1
230
Sick performance though!
7
I feel you bruh
my GitHub Actions spend is going out of control
55
Opus 5.5 is good
UPDATE: Opus 5.5 is good. Really good. Passes the vibe ✅
75
this is sofuckingood
The race for AGI Script: Sherpa by Pocket FM Video: Seedance 2.5
30
Opus 5.5 is OUT!
Claude Opus 5.5 is available today.
1
48
Grok 4.7 impressive stats
Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.
33
Call it what it is: lock-in. A connector built for Muse only works inside Muse. No other agent can touch it. This is App Store-era bullshit that should not exist anymore. Build on MCP and one connector works everywhere. Open always wins
Attention developers! You can now build your own Muse connectors.
1
1
115
Custom connectors are a good step. Now make them portable. If I build a connector, it should work with every agent I use, not just one company's bot. That's what MCP is for. Open beats walled garden
Attention developers! You can now build your own Muse connectors.
38
I'm back.
23,920
76,091
7,247
479,362
51,616,694
What shipped today: src/exemplar/adapters/typesafe.py, a new operator-registered reference adapter with two operations: •typesafe.judge - ask Jev any typed questions over state, get the probabilities back untouched. • typesafe.verdict - the decision layer. Claim plus evidence in, deterministic verdict out: proceed, hold, or insufficient_evidence. It asks Jev three calibrated questions (does evidence support the claim, does anything contradict it, what's the verdict), then derives the verdict in pure code under an operator-configurable policy. Every output records the full probability distribution, the policy applied, and the reason, so any verdict is auditable. @typesafeai
4 labs shipped new models in one week and nobody can keep up. Meanwhile nobody's testing the agents built on them. Exemplar: 4 packs that grade your agents like a strict QA lead. Free test: zaydreamer5.gumroad.com/l/ex…
3
1
1
126
exactly. judge hands you raw probabilities back, untouched. verdict wraps them in your policy and records the full distribution, so the decision is auditable. 2 jobs, 2 ops (this should be a name of a band btw)
17