Motion Drive retweeted
Replying to @boltdotnew @ShakibHimu
There were way too many great builds to pick just three. Shoutout to these 7 for making the judging extra hard
Enough: enough-flu9.bolt.host/
Partwith: partwith.app/
Keepsake Oils: keepsake-oils-web-ap-q0ud.bo…
FlightSuite: flightsuitehq.com
Scopesnap: scopesnap.bolt.host/
LuxePack 3D: luxepack-studio-qq6y.bolt.ho…
WrappedLedger: bank-statement-cinem-9ipm.bo…
Your turn. Start building 👉️ bolt.new
If your interface respects the rules of a pre-digital movement, it’s Punk.
My Decopunk service cards.
#illustrator
Réalisé à partir d'une photo de 1908 des frères Farman.
3d avec @MeshyAI
Motion Drive retweeted
Ray 3.2 Modify from my virtual motion control app I’m building with @grok’s help.
Created with Luma
@LumaLabsAI
Blue screen plate from virtual motion control app built with Grok to Ray 3.2 workflow. Reference image generated from extracted first frame by Luma and run with GPT 2.5 Sunburst. Stitching labels and music all orchestrated by Luma. Everything was done through conversation with the agent, no “prompts” were used.
#LumaCPP
🚨🇫🇷💻 ALERTE INFO | Les données fiscales de plusieurs centaines de milliers de Français pourraient avoir été DÉROBÉES lors d’une cyberattaque.
Un hacker revendique avoir infiltré des serveurs internes de la Direction générale des Finances publiques (DGFiP) et exfiltré 678 438 lignes de données avant que son accès ne soit coupé. La DGFiP n’a pas encore confirmé l’information. (French Breaches)
Motion Drive retweeted
I tested Isaac (Pokee) on a real use case.
A music studio: intention → finished track.
Style, vibe, duration..
Two minutes later, the audio is ready.
Isaac handles analysis and planning end-to-end.
10 million tokens of context.
No chunking. No chatbot.
Full track analysis: 1 cent.
The system runs autonomously all the way to display the result on the player.
@threejs @MeshyAI @Pokee_AI
#Pokee #AIAgent #music
Meta just released Muse Glimmer 30B. So we put it side-by-side with Pokee-Isaac 28B, Qwen3.6 27B, and Gemma 4 31B.
Same ~30B weight class. Same evaluation environment.
Pokee-Isaac leads across every agentic benchmark we tested:
BFCL v4: 70.94
τ³-bench: 66.67
Terminal-Bench 2.1: 65.12
MCP-Atlas: 74.59%
And then there’s Pokee-Isaac's unique long-context capabilities:
93.3% RULER at 10M context
0.5 MRCR v2 at 1M token context
where all other models score 0!
All in a 28B model deployable on a single GPU.
The ~30B local frontier is getting very interesting.
I tested Isaac (Pokee) on a real use case.
A music studio: intention → finished track.
Style, vibe, duration..
Two minutes later, the audio is ready.
Isaac handles analysis and planning end-to-end.
10 million tokens of context.
No chunking. No chatbot.
Full track analysis: 1 cent.
The system runs autonomously all the way to display the result on the player.
@threejs @MeshyAI @Pokee_AI
#Pokee #AIAgent #music
Releasing Pokee-Isaac 28B — the world’s first real 10M-token context frontier-class agentic model, deployable on a single GPU (starting from RTX 4090 or equivalent).
New proprietary non-decoder-only architecture:
• 93.3% RULER at 10M tokens
• Up to 137K tokens/s prefill on one B200 with 10M-token context
• Leads BFCL v4 and τ³-bench in our evaluation
• Lowest combined attack success rate among evaluated models on DTAP security red-teaming benchmark
Pricing and deployment:
💰 $0.15/M input · $1/M output
🔒 Deploy in your VPC, on-premises, or on-device, with Day-0 support for @vllm_project and @sgl_project
Technical blog: console.pokee.ai/model
Technical report: console.pokee.ai/pokee-isaac…
API: console.pokee.ai








