David Brzozowski

passion for building things CTO @medacist (acquired by @bluesight_inc) pursuit is the evidence of desire

[OpenClaw🦞 @openclaw]

Grok 4.5 from @SpaceXAI is live on OpenClaw.

No OpenClaw update required, just connect your X Premium or SuperGrok subscription, select Grok 4.5 under the xAI provider, and use an Opus-class model that’s fast, low cost, and ready for agentic work.

[David Brzozowski @davidbrzozowski]

When making a prediction about the future, favor your own intuition over validation from peers

[David Brzozowski @davidbrzozowski]

🚀🇺🇸

[David Brzozowski @davidbrzozowski]

Historically META has used Everpure (frmly Pure Storage) as it's exclusive storage provider in it's datacenter builds. If they continue to build on that relationship, this will be a catalyst for Everpure

$META $P

[David Brzozowski @davidbrzozowski]

@Jason @Ubiquiti I don't know of a hardware vendor that innovates with the speed and breadth of @Ubiquiti

[David Brzozowski @davidbrzozowski]

The model commoditization complex

[Claude @claudeai]

Sonnet 5 is a substantial improvement over Sonnet 4.6 on reasoning, tool use, coding, and knowledge work.

Its performance is close to Opus 4.8, at lower prices.

[Sebastian Raschka @rasbt]

Have been taking different local open-weight LLMs for a test drive in different harnesses (Qwen-Code, Codex, Claude Code).

30B Mixture-of-Expert models are kind of a nice sweet spot and can solve challenging problems. And they get roughly 40 tok/sec on a Mac or DGX Spark, which is similar to GPT 5.5 in a Pro subscription and totally useable for everyday work.

More interesting is also the harness choice! Claude Code seems to be using 2x many tokens as Codex.

Gemma 4 E2B is here just for reference to show that the tasks can't be trivially solved by smaller models.

Just finishing a longer write-up about this and will share soon (likely tomorrow)!

[David Brzozowski @davidbrzozowski]

Pursuit is the evidence of desire

[David Brzozowski @davidbrzozowski]

of 61 iOS apps that needed updating on my phone ONLY 24 of them included detailed release notes

kudos to @github for sharing what’s packaged

[Anthropic @AnthropicAI]

We’ve received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5.

We'll begin restoring access tomorrow, and will share an update soon.

We’re grateful to our users for their patience, and to everyone who worked with us on redeploying the models.

[David Brzozowski @davidbrzozowski]

for low latency high bandwidth cluster wide (across several racks) decode to be feasible, fiber/photonics have to live at the interconnect layer. Copper is too slow and distance limited (important for links longer than 3meters)

[Etched @Etched]

Vertically integrating the full inference hardware stack is extremely hard.

Hear @UbertiGavin and @robertwachen share three years of battle stories leading into today's launch

[Shaan Puri @ShaanVP]

if you simply:

- dont bet on sports
- dont install instagram/tiktok on your phone
- dont eat fast food

you’re probably in the top 1% of the population

success isnt as much about being smart as it is about just not being stupid

[Barchart @Barchart]

Bitcoin $BTC closed below its 200-week moving average for the first time since 2023 🚨 This has historically been a great buying opportunity 🤯 👀

[David Brzozowski @davidbrzozowski]

@DeepValueBagger Whether they are using OSS models or U.S. Frontier models, inference is inference. The demand is real.

[Peter H. Diamandis, MD @PeterDiamandis]

The world is full of people predicting the future and short on people building a piece of it. Prediction is a spectator sport.

[David Brzozowski @davidbrzozowski]

@aleabitoreddit I like $APH for this reason..
Catalyst’s now:
- Strong FCF to fund expansion
- AI Buildout (chips are current constraint, connectivity is the next)
- Energy storage and transmission
Future Growth:
- Defense Industry (autonomous is the future)
- Space
- Robotics

[Jubayer Ibn Hamid @jubayer_hamid]

The most capable reasoning systems in AI scale inference compute along several axes: sequential compute to think longer, parallel compute to sample many independent attempts, and aggregative compute to synthesize prior traces into a new improved one. But during training, we only optimize how models use sequential compute. This creates a fundamental mismatch between how we ultimately deploy these systems and how we train them, leaving much of search and synthesis unoptimized.

We introduce SPIRAL, an RL framework for making all inference-compute primitives end-to-end learnable: models learn to coordinate sequential, parallel, and aggregative reasoning using only the reward of the final output. Work with @ifdita_hasan (co-lead), @michaelyli_ , @oshaikh13 , @yoonholeee , @DorsaSadigh , @chelseabfinn , @noahdgoodman 🧵

[BlackRock @BlackRock]

Bitcoin’s role in portfolios is evolving, and it could be considered a complementary diversifier.

We believe a modest allocation (typically ~1–2%) could impact return potential in a portfolio while maintaining appropriate risk tolerance.

Hear more from Michael Gates on how bitcoin’s unique properties may complement traditional portfolios ⏩ https://t.co/hydyHrgFi5

#FinancialAdvisors #Cryptocurrency #Bitcoin

[David Brzozowski @davidbrzozowski]

Inline metadata is a game changer for agent consumption

[Mistral AI @MistralAI]

Introducing Mistral OCR 4. It creates structure with bounding boxes, block classification, and inline confidence scores in 170 languages. 🧵👇

[Trump War Room @TrumpWarRoom]

The European mind is starting to comprehend. 🇺🇸