Al Enthusiast | Ghostwriter | Empowering buisness models & personal brands to grow with latest tech trends. DM for collab
California, USA
Joined March 2024
- Tweets14K
- Following6.3K
- Followers14K
- Likes14K
The Self-Improvement Loop
Ornith-1.5 isn't just learning from data.
It can generate tasks, create scaffolds, produce rollouts, and turn them into new training experiences.
AI isn't just being trained anymore.
It's increasingly helping shape its own training.
Aloha! 🌺Introducing Ornith-1.5, a family of open-source LLMs spanning 9B Dense, 35B MoE, and 397B MoE, trained with self-improving strategies.
It achieves state-of-the-art performance among open-source models of comparable size and delivers performance comparable to Claude Opus 4.8 across reasoning, agentic, and coding tasks:
✅Terminal-Bench 2.1 (86.1)
✅SWE-Bench (86 on verified, 65.1 on pro, 79.6 on Multilingual)
✅DeepSWE (56)
✅HLE (44.6)
✅ClawEval (81.4)
✅Tool Decathlon (71.2)
Ornith-1.5 takes a major step toward training foundation models through end-to-end self-improvement, extending the self-scaffolding strategies introduced in Ornith-1.0 into a more complete self-improvement loop: the model proposes new tasks, generates task-specific scaffolds, and produces solution rollouts for reinforcement learning, continuously creating new learning experiences from which it can improve.
All models, along with their quantized versions (FP8, GGUF, MLX, and NVFP4), have been released under the MIT License, enabling unrestricted commercial and research use.
📘Tech Blog: ornith.ai/ornith_1_5.html
🤗Huggingface: huggingface.co/collections/o…
The AI race is moving from biggest models to best systems. Pokee‑Isaac 28B backs it up with a true 10M‑token context, frontier‑level agents, enterprise deployment, and pricing that enables scale.
🤝 Paid partnership
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
We're witnessing the birth of a new creative medium. Grok thinks, HyperFrames visualizes, and ideas become cinematic experiences in minutes. That's a profound shift. 🚀
🤝 Paid partnership
HyperFrames is now officially live on @Grok
Grok knows everything. HyperFrames shows it
Any answer becomes a polished video in minutes
Proof: this video was created entirely with it
The creator economy rewards speed and consistency.
A tool that cuts editing time while improving content quality gives creators a significant advantage from day one. 📈
🤝 Paid partnership
the boring backend problem is always the most important one @emergentlabs 📈 and you guys actually went after it
1/ @RCBTweets winning again. Rajat Patidar at the center of it
And now Rajat coming back to @MPLeagueT20, which starts today. Cricket has had a beautiful week
But what @AScindia is building for MPL on @emergentlabs is the part I am excited and inspired about the most
What stands out about MuleRun Studio is the shift from generation → control.
Most AI tools still feel unpredictable.
This feels structured, editable, and production-oriented.
That’s the difference between a fun demo
and software creators actually build businesses on.
🤝 Paid partnership
Most “AI workflows” still feel stitched together.
This feels different.
Everything happening inside one workspace changes the entire speed of delivery.
🤝 Paid partnership
The funniest part of this whole analysis is the UK turning Mother’s Day into the stationery Olympics
A heartfelt card beats an expensive gift.
🤝 Paid partnership
A lot of models look impressive in demos.
Very few are designed for nonstop real-world execution under production load.
Ring-2.6-1T leaning hard into agent reliability and adaptive compute feels like a very smart bet 🔥
And yes it’s free for a week.
🤝 Paid partnership
We are launching Ring-2.6-1T, a trillion-parameter flagship thinking model engineered for real-world complex tasks and production env: 🚀
- Adjustable Thinking Effort: dynamic compute mechanism to flexibly balance cognitive depth, token cost, and execution speed;
- Agent-Optimized: Built for high-frequency workflows, delivering rapid multi-step execution and tool orchestration with SOTA stability;
- Deep Thinking: Unlocks the model's maximum capability ceiling for rigorous mathematical logic and scientific research;
“Where is Kris?” a full “Where’s Waldo”-style crowd search poster set in New York City.
Broadway signs, yellow cabs, Wall Street hundreds of tiny characters, all from one GPT-Image-2 prompt.
Now live on MuleRun. The creative prompts in this collection keep surprising.
🤝 Paid partnership
This quietly changes how content gets made.
One system. One flow. Zero handoffs.
Claude handles thinking, Pixa handles doing.
You just hit publish more often.
We didn’t just build AI to solve problems.
We built systems that discover better ways to think about problems than we ever could.
The last stronghold of coding has just been conquered by AI.
In the most recent three Codeforces live competitions, i.e., Round 1087, Round 1088, and Round 1089, GrandCode, our agentic AI system, ranked first in all of them, beating all human participants, including legendary grandmasters.
GrandCode is a multi-agent reinforcement learning system designed for competitive programming. It orchestrates a variety of agentic modules (hypothesis proposal, solver, test generator, summarization, etc) and jointly improves them through post-training and online test-time RL. GrandCode is developed based on Qwen. Huge respect to the Qwen @Alibaba_Qwen team for their contributions to the community.
It is hard to imagine how quickly AI has advanced in just one year:
1st — GrandCode (March 2026)
8th — Gemini 3.1 Pro (February 2026)
175th — OpenAI o3 (April 2025)
We can’t wait to see what happens over the next year.
Readers added context they thought people might want to know
Relying on AI to generate algorithmic logic or the key solution is strictly prohibited.
codeforces.com/blog/entry/133…
Apparently Spotify and Apple teams use Willow Voice?? The Teams launch just made this the easiest call for any company that actually cares about how their people work. Congrats on the launch!
Seedance 2.0 fixes short-clip pain: multi-shot + consistent characters for seamless films—powered by Elser AI’s film-engine workflow.
UNLIMITED VIP (2 years): 87% OFF top-tier (Kling 3, Nano Banana Pro, Midjourney V7, Veo 3.1).
@ElserAIOfficial
#Seedance2 #ElserAI
🎬 Seedance 2.0 is coming.
Elser AI is the world's first studio-grade AI film engine and Seedance 2.0 makes it even more powerful. ⚙️
🎞 True multi-shot storytelling
🎥 Built for real AI filmmaking
🚀 Coming soon.
🎟 Early access available — link in replies 👇
#Seedance2 #ElserAI #AIFilmmaking #AIVideo
Vibe coding = describing, not typing. Big shift.
Story Mode structures the narrative, Agent Opus perfects the presentation.
Together they transform raw ideas into refined videos that actually flow. With free credits available, it’s the ideal time to test your most ambitious concepts.
"Sometimes I'll start a sentence, and I don't even know where it's going. I just hope I find it along the way." — Michael Scott & every AI video generator (until now).
Today, we’re launching the greatest storytelling tool for video: Story Mode.
To celebrate we're giving everyone FREE CREDITS. Just sign up to get started on your next viral masterpiece.
Try it while it's free 👇
Style inconsistency has been the silent killer of AI video quality.
You could generate amazing shots, but they never felt connected.
Agent Opus Story Mode finally treats a video like one cohesive film instead of random clips stitched together.
The biggest problem in AI video is style inconsistency - every scene has different styles and transitions are "jumpy".
That's why we're launching Story Mode in Agent Opus.
Generate style-consistent videos up to 4 mins long in just one click.
✅ Your original audio or script
✅ Your custom style reference
✅ 16:9 | 9:16 | 1:1 formats ready for monetization
No more random, glitchy scenes. Just one continuous, high-retention shot.
Try it today for free 👇