Trilogy AI Center of Excellence retweeted
Kimi K2.5 is beating Claude Opus 4.5 and GPT-5.2 on key benchmarks at 1/5th to 1/10th the cost.
Here's how to route it through Fireworks via OpenRouter, plus configs for Claude Code, OpenHands, and Droid.
Article & video links in comments.
🥝 Meet Kimi K2.5, Open-Source Visual Agentic Intelligence.
🔹 Global SOTA on Agentic Benchmarks: HLE full set (50.2%), BrowseComp (74.9%)
🔹 Open-source SOTA on Vision and Coding: MMMU Pro (78.5%), VideoMMMU (86.6%), SWE-bench Verified (76.8%)
🔹 Code with Taste: turn chats, images & videos into aesthetic websites with expressive motion.
🔹 Agent Swarm (Beta): self-directed agents working in parallel, at scale. Up to 100 sub-agents, 1,500 tool calls, 4.5× faster compared with single-agent setup.
-
🥝 K2.5 is now live on kimi.com in chat mode and agent mode.
🥝 K2.5 Agent Swarm in beta for high-tier users.
🥝 For production-grade coding, you can pair K2.5 with Kimi Code: kimi.com/code
-
🔗 API: platform.moonshot.ai
🔗 Tech blog: kimi.com/blogs/kimi-k2-5.htm…
🔗 Weights & code: huggingface.co/moonshotai/Ki…
Trilogy AI Center of Excellence retweeted
I tested 4 "agentic retrieval" systems against a simple dense baseline.
The results? Every single "advanced" system performed worse.
Here's what several hundreds of $$ in API calls taught me about the agentic retrieval hype
Trilogy AI Center of Excellence retweeted
The full study is here - trilogyai.substack.com/p/age…
Trilogy AI Center of Excellence retweeted
We stand at a fundamental fork in the road for software development.
After analyzing 11 leading #AI coding agents, one truth emerged: the real revolution isn't about autonomous agents, but autonomous developers.
A thread on the choice that will define our future 🧵
Trilogy AI Center of Excellence retweeted
🎥Text-to-Video AI is evolving at breakneck speed, transforming from simple clips to sophisticated videos with realistic physics and complex narratives.
But there's a massive gap between what's possible and what's practical. Let me break down the state of the art:
Trilogy AI Center of Excellence retweeted
I'm excited to be giving a workshop at Convergence 2025!
Navigating Uncharted Metrics:
Reference-Free LLM Evaluation with G-Eval
🧠 Learn how to benchmark LLMs using rubric-based evaluation.
🕒 May 14, 3:05-4:05PM ET
🔗 comet.com/site/events/conver…
#ConvergenceConference @Cometml
🎉 We’re thrilled to announce the launch of the Trilogy AI Center of Excellence website!
Explore cutting-edge #AI research on topics like multi-agent systems, and access tools like our Reference-Free LLM Evaluation Framework.
Dive in now! ai-coe.trilogy.com/
🚀 Just launched: The Trilogy AI Center of Excellence website is now live!
I've created a central hub to organize all our #AI research, publications, and resources.
Still evolving, but too valuable to keep under wraps any longer.
Explore it here: ai-coe.trilogy.com/






