Reaching for the stars! 🌌🚀📡
Houston
Joined September 2022
- Tweets3.1K
- Following398
- Followers28
- Likes13.6K
JuliaSpaceVision retweeted
Nemotron 3.5 Super VL from @nvidia is now available on Modal.
With 120B total parameters and 12B active per token, this open-weight multimodal model brings powerful visual reasoning to documents, images, and application screenshots—enabling sophisticated agentic workflows.
JuliaSpaceVision retweeted
A new open-source release cuts MiniMax H3's generation from 49 steps to 8.
It's called HyperFlow. In Video Rebirth's H200 tests, clips finished about 3X faster, with better camera control and more detail.
Three workflows, with the demos below ↓
JuliaSpaceVision retweeted
CRASH ZOOM
One zoom. Instant panic.
A crash zoom forces your audience onto the detail that changes everything.
Don’t spend it on a random face.
Here’s a 20-second setup built around that moment.
Prompt below 👇
#DreaminaCPP
@dreamina_ai
JuliaSpaceVision retweeted
YOU DON'T HAVE TO PICK ONE AI MODEL.
Cline is an open source coding agent that works with ANY of them.
Claude.
GPT.
Gemini.
Local models on your own machine.
It edits multiple files, runs commands, and uses the browser.
Inside VS Code or straight from the terminal.
Bring your own key.
Switch models whenever you want.
70k+ GitHub stars. Zero lock in.
github.com/cline/cline
JuliaSpaceVision retweeted
Seven ways to encourage critical thinking in #AI users
buff.ly/lKBcxZf
#business #leadership #management #organization #tech #HR #CHRO #CIO #CTO #CDO #CEO #digital #innovation #disruption #digitaltransformation #automation #genAI #AgenticAI #skills #talent #talentmanagement #jobs #careers #futureofwork @InfoWorld @NYIke @jblefevre60 @pierrepinna @kalydeoo @Ym78200 @CurieuxExplorer @mvollmer1 @JoannMoretti @anand_narang @EvanKirstel @SpirosMargaris @ipfconline1 @Jeffkagan @AndresVilarino @efipm @FMFrancoise @MHcommunicate @JeromeMONANGE @mary_gambara @drsharwood @smoothsale @debashis_dutta @devbhatt72 @MarshaCollier @PamMktgNut @arlenenewbigg @JagersbergKnut @TysonLester @GlenGilmore @pchamard @baski_LA @AnthonyRochand @NewsNeus @domingonarvaez1 @bamitav @jornalistavitor @c4trends @amalmerzouk @ralf_ladner @netzpalaver @nafisalam @cto_software @Oschonschek @Matt_Rosenquist @Shirastweet @Kevin_Jackson @JimMarous @chidambara09
JuliaSpaceVision retweeted
Radiation Limits On Amino Acid Preservation And Chiral Measurement At Oxia Planum
astrobiology.com/2026/10/16/… #astrobiology #Mars #biosignatures
JuliaSpaceVision retweeted
I've seen this happen many times in recent months. When embracing AI, people who are really good at their craft become even better.
Experts who don't adjust to AI agents will be left behind. And non-experts who think they don't need to learn anything because AI agents can do everything will also hit obstacles.
Having domain expertise makes a huge difference when instructing and steering AI agents.
JuliaSpaceVision retweeted
Understanding a video means comparing moments, and models are getting much better at it.
Glimpse from @ElorianAI checks how a scene changes from one point in time to another, and goes back for a closer look when the change is small or fast.
Elorian calls this visual thinking: reasoning over what the model observes, grounded in the frames.
Glimpse can return to the source during reasoning, crop into an area, extract specific video frames or run code to measure or compare what it sees, a mechanism Elorian calls a “visual scratch space.”
🤝 Paid partnership
JuliaSpaceVision retweeted
people that don't think local inference will be a thing are crazy
every computer is going to have a series of small models running for latency/privacy
JuliaSpaceVision retweeted
NASA Exoplanet Explorers (ExoExplorers) Applications Due on Thursday 15 October
astrobiology.com/2026/10/15/… #astrobiology #exoplanet #astronomy @NASAScience_
JuliaSpaceVision retweeted
Moondream 3 Preview and Moondream 3.1 are live on OpenRouter
Small vision-language models (9B total, 2B active MoE): image + text in, 32K context, optional reasoning.
$0.30/M input tokens
$2.50/M output tokens
nitter.cf/OpenRouter/status/2110…
Moondream 3 Preview and Moondream 3.1 from @moondreamai are now on OpenRouter, served by @PrimeIntellect
Object detection and pointing work in plain chat: attach one image and start your message with /detect {"object":"coffee cup"} or /point {"object":"croissant"}
JuliaSpaceVision retweeted
How can AI learn to do better research?
I’ll explore memory, collaboration and evolutionary search in my keynote, “Learning to Discover: The Path to Self-Improving AI,” at DEEP DIVE INTO 2026 conference
Seoul · Nov 10–12
deepdiveinto.kr/ – at Minato-ku, Tokyo
JuliaSpaceVision retweeted
Hy Memory 🧠 New memory model for AI agents from @TencentHunyuan
- 30B MoE / 3B active
- Apache 2.0
- Extracts important memories from conversations
- Updates outdated memories automatically
JuliaSpaceVision retweeted
Never ever doom, you can be upset and frustrated and angry, and vocal about your discontent. But never doom.
I see an impossible tech job market, I pivot to education in the meantime. Pays the bills. Gets me out of the house. Keeps me grounded. Sure the pay is a fraction of what you make in tech, but the game isn’t about economically thriving right now. It’s about staying alive.
Life is beautiful. Everyday is a blessing. Never be afraid to ask for help.
JuliaSpaceVision retweeted
As GraphRAG systems grow, not every decision needs to go through an LLM. Partha Sarkar looks at how TypeSafe Jev can handle frequent graph decisions while LLMs focus on the tasks that require deeper reasoning.
towardsdatascience.com/graph…
JuliaSpaceVision retweeted
Google achieved a new level in quantum computing.
Says it has cleared a bottleneck researchers have flagged since 2004.
Quantum computers need single-use “magic states” to run the operations that put them beyond ordinary computers.
Making clean ones is so costly that some estimates say it could take up to 95% of a machine’s capacity.
Google’s new method checks each state and throws out flawed attempts, cutting errors about 40x on its 105-qubit Willow chip.
The surviving states reached 99.99% accuracy, a record by Google’s count, in work published in Science today.
Only 8% of attempts survived the filter, and the method still has to work at much larger scale.
Google has shown on real hardware 1 of the methods behind its estimate that RSA-2048 encryption could be broken with under 1M qubits.
Some magical progress towards "magic state" in quantum!
Our @GoogleQuantumAI team has demonstrated "magic state cultivation" on physical hardware for the first time. The result is a 40x error reduction on our Willow processor vs. previous methods, achieving a state-of-the-art 99.99% logical fidelity.
It's another step towards useful, error-corrected quantum computers to help solve the hardest problems. The result is published in Science today, watch the video below to see how it works
JuliaSpaceVision retweeted
This is the kind of 30-second shot that proves camera direction matters more than explosions. 👀
The entire concept lives or dies on one move:
hallway → peephole → apartment → peephole → hallway
No cut. No breach shown. Just tension building on both sides of the same door.
That’s exactly the kind of cinematic structure Seedance 2.5 can make ridiculously effective.
Full prompt below 👇
#DreaminaCPP
@dreamina_ai
JuliaSpaceVision retweeted
FLUX 3 Image Klein from @bfl_ai can:
- Run on a single consumer GPU with 5.6B params
- Generate at native 4K
- Place things exactly with bounding boxes
- Do pixel-perfect edits
- Edit with multiple references
- Make an image in ~4s on an RTX 5090 (INT8)
Tested it myself, here's the result 👇
Prompt:
A cinematic high-fashion photograph treated with the texture and lighting of a 19th-century oil portrait, featuring a young East Asian woman standing in a dark, ornate penthouse at night. She wears a structured black silk evening gown with a high collar and long sleeves, the fabric catching light along its sharp folds and open back. One hand rests on a tall window frame as she looks outward over a glittering city. Cool moonlight from the side mixes with warm interior light, creating strong chiaroscuro across her face, neck, and the gown’s silhouette. Shallow depth of field, visible film grain, rich color grading with deep shadows and selective gold highlights. Intimate yet distant mood, refined and quietly captivating. aspect_ratio:16:9 resolution:2k
🤖 Made with AI
JuliaSpaceVision retweeted
Closed-source sky has fallen...
these 10 GitHub REA repos let your AI agent reverse engineer any app, game or tool on any device
send to your Claude:
01. REA (130K stars) - one MCP for RE across binaries, Electron, .NET, Android and websites.
▸ github.com/morluto/rea
02. Ghidra (84K stars) - NSA's free decompiler. The engine under most AI RE setups.
▸ github.com/NationalSecurityA…
03. ImHex (55K stars) - hex editor with an MCP server for agent file inspection.
▸ github.com/WerWolv/ImHex
04. mitmproxy (45K stars) - intercepts traffic and maps the hidden APIs behind it.
▸ github.com/mitmproxy/mitmpro…
05. ILSpy (27K stars) - .NET back to readable C#.
▸ github.com/icsharpcode/ILSpy
06. Frida (22K stars) - runtime hooking on desktop and mobile. CLI-scriptable.
▸ github.com/frida/frida
07. GhidraMCP (11K stars) - the agent bridge. Headless or GUI.
▸ github.com/LaurieWired/Ghidr…
08. Android RE Skill (8K stars) - APK to decompiled code plus API endpoints.
▸ github.com/SimoneAvogadro/an…
09. Universal Modder (7K stars) - agent skills for modding PC games.
▸ github.com/rehan-remade/univ…
10. Cpp2IL (3K stars) - Unity C++ back to .NET for ILSpy. WIP.
▸ github.com/SamboyCoding/Cpp2…
Paste this into your Claude, then read how to reverse engineer anything with Opus 5.5 below.
