frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Show HN: InstinctFlash – Run 5B world-action models in real time on Jetson Thor

https://github.com/General-Instinct/InstinctFlash
8•guanming0717•2h ago
Hey HN, Guanming here, cofounder of General Instinct. We just released InstinctFlash, a high-performance serving framework for robotics models. It’s licensed under AGPL-3.0.

On Jetson Thor, we see speedups about 1.2x to 7.9x from runtime optimizations alone and up to 33.78x for LingBot-VA when we combine those runtime optimizations with a distilled few-step diffusion scheduler, going from the original 25 visual / 50 action steps to 2 / 4 steps. Across 50 Robotwin2.0 tasks, we evaluated 1,153 episodes per configuration, LingBot-VA with InstinctFlash at 2 visual / 4 action steps achieved a 90.5% success rate, compared with 92.1% for the baseline at 25 visual / 50 action steps.

Here’s an optimized 5B world action model, running in real time on a Jetson Thor: https://youtu.be/nku65iyL5Fw

InstinctFlash currently supports 8 VLA / world-action model families, including pi0.5 and NVIDIA Cosmos Policy, across RTX 4090 / 5090 and Jetson Thor.

Just give it your fine-tuned checkpoint and InstinctFlash handles the rest, exposing the accelerated model through a Python runtime or an OpenPI-compatible WebSocket server.

We started working on this because we kept running into the same problem while deploying robot policies, the models were getting much better, but inference was often way too slow for the control loop we actually wanted.

For pi0.5, mixed-precision GEMMs and CUDA graphs speed up computation and reduce launch overhead. For Cosmos, caching avoids redundant computation across diffusion steps. World-action models’ diffusion denoising step depends on the previous one which motivated our work on few-step distillation.

Right now, InstinctFlash contains 6 aspects of optimization.

- Graph: CUDA graph capture, memory planning and separating prefill from repeated execution.

- Cache: Reusing KV and conditioning state across diffusion steps and prediction calls.

- Attention: Specialized attention paths for different model architectures.

- Kernels: Fused operations and kernels tailored to specific backends and tensor layouts.

- Precision: FP8 and mixed-precision execution.

- Model: Few-step distillation for diffusion and action generation.

Teams at Samsung, Siemens, and other robotics startups have used InstinctFlash for model acceleration on VLAs, WAMs, and diffusion-based world models. Now we are opening up access to you.

Feel free to try it here: https://github.com/General-Instinct/InstinctFlash

More implementation details and benchmarks: https://general-instinct.com/blog/instinctflash-edge-inferen...

Would love to hear your feedback!

Claude Opus 5.5

https://www.anthropic.com/claude-opus-5-5
335•km144•1h ago•401 comments

OpenAI GPT–6 Astra breaks Enigma message that has resisted solution since 2005

https://www.cryptocellar.org/bgac/the-mvueh-break.html
354•sohkamyung•3h ago•279 comments

There's a high chance of devices being sold with GrapheneOS preinstalled in 2027

https://grapheneos.social/@GrapheneOS/117299954135808210
35•Cider9986•34m ago•7 comments

Claude Opus 5.5 Intelligence, Performance and Price Analysis

https://artificialanalysis.ai/models/claude-opus-5-5
43•theanonymousone•56m ago•22 comments

WordPress: Unauthenticated path traversal leading to conditional RCE

https://github.com/WordPress/wordpress-develop/security/advisories/GHSA-7hp8-65ch-5whp
36•vntok•1h ago•15 comments

16-bit Intel 8088 chip

https://allpoetry.com/16-bit-Intel-8088-chip
53•rbanffy•1h ago•7 comments

OpenAI is well positioned to fast-follow Jev

https://arcturus-labs.com/blog/2026/09/21/will-openai-eat-jevs-lunch/
158•JohnBerryman•3h ago•113 comments

Launch HN: Coverage Cat (YC S22) – Umbrella insurance via your personal agent

https://www.coveragecat.com/
8•botacode•21m ago•3 comments

Writing Rust code that's fast by asking agents to make the code faster

https://minimaxir.com/2026/09/agentic-iteration/
50•mooreds•2h ago•21 comments

Apple has added persistent 'ads' to iOS, and it's driving users crazy

https://www.techradar.com/phones/iphone/i-wish-apple-would-just-stop-that-crap-apple-has-added-pe...
377•MC995•3h ago•282 comments

Show HN: Drop – A rootless Linux sandbox with gVisor support

https://droprun.sh/
106•mixedbit•3h ago•34 comments

Show HN: AI·rete·RAG – a Rete rule engine decides, RAG explains why

https://ai-rete-rag.com/
14•ZaharaHussain•1h ago•0 comments

Solitaire Alone Together

https://solitairealonetogether.com/
71•eieio•19h ago•19 comments

Can gzip be a language model?

https://nathan.rs/posts/gzip-lm/
333•networked•11h ago•127 comments

AMD's random number generator can't generate a 0?

https://board.flatassembler.net/topic.php?t=24261
203•BruceEel•9h ago•154 comments

MiMo v2.6

https://mimo.xiaomi.com/mimo-v2-6
1051•volf_•21h ago•462 comments

One Minute Park

https://oneminutepark.tv/
8•namuorg•1d ago•3 comments

Show HN: InstinctFlash – Run 5B world-action models in real time on Jetson Thor

https://github.com/General-Instinct/InstinctFlash
8•guanming0717•2h ago•0 comments

Spymarks, not Watermarks

https://brand.io/article/spymarks/
619•possibilistic•18h ago•156 comments

Meta’s Muse has a serious 0-day

https://arstechnica.com/security/2026/09/muse-metas-extraordinarily-privileged-ai-assistant-has-a...
80•pavel_lishin•3h ago•32 comments

Teleoperated Humans

https://www.jefftk.com/p/teleoperated-humans
33•luu•2d ago•37 comments

I asked Meta’s Muse for its filesystem and it sent me 6.8GB

https://mouse.dev/blog/muse-runtime-export/
194•Aeroi•2h ago•106 comments

The Economics of Open-Weight Inference

https://data.ornn.com/publications/the-economics-of-open-weight-inference
20•marinesebastian•4h ago•9 comments

MUNI Heritage Weekend in San Francisco

https://daniel.lawrence.lu/blog/2026-09-20-muni-heritage-weekend/
106•plun9•1d ago•21 comments

Transformers Explained Visually

https://poloclub.github.io/transformer-explainer/
572•aray07•22h ago•84 comments

Relativistic raytracing

https://publish.obsidian.md/h1m3/Articles/Relativistic+raytracing
14•vismit2000•1d ago•2 comments

Vacate a drone restriction that criminalized recording immigration agents

https://www.eff.org/deeplinks/2026/09/dc-circuit-must-vacate-drone-flight-restriction-criminalize...
57•hn_acker•2h ago•12 comments

I said no and Apple said yes

https://dbushell.com/2026/09/22/apple-intelligence/
675•thatslast•9h ago•543 comments

I don't want to read what you didn't write

https://blog.colinbreck.com/i-dont-want-to-read-what-you-didnt-write/
952•mooreds•19h ago•392 comments

MiMo-v2.6-Pro: Intelligence, Performance and Price Analysis

https://artificialanalysis.ai/models/mimo-v2-6-pro
140•theanonymousone•13h ago•60 comments