🚨 FUSION AGENTS - Build Complete SaaS Apps With Open-Source AI Fusion agents combine Kimi 2.7 and GLM with Opus 4.8 and GPT 5.5 - multi-agent architectures with open-source sub-agents - build complete SaaS apps - one click connectors to 100 products - create companion iOS and android apps with one prompt - accept stripe payments

🚨Anthropic flew senior security team to Washington to meet with Commerce and National Cyber Director Reporter: How long until this is resolved? WH official: “That’s up to Anthropic.” Another official warned: “The longer this goes on, the more likely it becomes a de facto licensing regime.” “Every model going forward needs to ask the government’s permission for whether it can be released.” “That’s an extremely bad situation.”

I spent the last 6 months building my own harness and orchestrator. I built it to allow me to experiment on the frontier of ideas. Little did I know that the orchestration, the harness, routing capabilities, dynamic artifacts/workflows, verifiers, ability to switch/route between agent backends, automations, the skills, and the MCP tools would be the absolute best defense for what happened with Fable this week. The argument folks made when I was talking about "owning the agent orchestrator" at the beginning of the year is that this is just high maintenance, too costly, and is unsustainable. It might still feel like it to many. But there is too much to lose if you decide to lock yourself in with a specific tool or model provider. Really, the way I have built my orchestrator is through mining my agent sessions and using that to recursively build and test our new ideas that range from autonomous loops to continual learning/memory systems. I can test research ideas on the fly. I just can't go back to using a vendor that only offers me a set of features. My argument now is that you really don't have a choice. You need to be able to control cost, decision making, context management, and everything in between. If you don't, then how are you going to tap into the world of recursive self-improving AI? It won't get any easier if you don't own the decision-making part of the intelligence stack.

We benchmarked 7 frontier models on 3 categories of autoresearch tasks: ML engineering, harness/prompt engineering, and algorithmic discovery. Fable-5 won overall even under cost constraint, but on ML engineering, the open model Kimi-K2.7-Code surpassed frontier models.🧵(1/5)

Why V4-Pro is so close to V4-Flash (which looks great for Flash and horrible for Pro) is one of the most important questions for the trajectory of open source. If I'm right and it's distilled from Flash experts, they will dominate. If this recipe isn't scalable… good for Dario.

SITUATION EXPLAINED: @ml_angelopoulos, CEO of @arena joined us to explain how Claude Fable 5 is performing. "What this means is that, compared with the average model, Claude Fable 5 is getting about 11% more positive signals from real agentic workflows. In the middle, models like GLM and Gemini 3.5 Flash are roughly at baseline, and then performance gets worse as you move down the leaderboard." "The signals on this leaderboard come from a few different sources. The first is task success rate: explicit signals where users say they were able to complete the task using a given model. Fable has an 18% task success rate. By comparison, the next-best competitor, Claude Opus 4.8 Thinking, is at 9%. So Fable almost doubles the effect of the next-best model. That’s a massive improvement in people’s ability to get the job done."