Open source Xiaomi MiMo-V2.5 and V2.5-Pro are among the
most efficient (and affordable) at agentic 'claw' tasks

Xiaomi, the Chinese firm best known for its smartphones and electric vehicles, has lately been shipping some incredibly affordable and high-powered open source AI large language models.

The trend continued today with the release of Xiaomi MiMo-V2.5 and Xiaomi MiMo-V2.5-Pro, both available under the permissive, enterprise-friendly MIT License, making them suitable for use in production in commercial applications.

The most notable attribute of these models besides the open source licensing is that, according to Xiaomi's published benchmarks, they are among the most efficient available for agentic "claw" tasks, that is, powering systems such as OpenClaw, NanoClaw and Hermes Agent, in which users can communicate with them directly over third-party messaging apps and have the agents go off and complete tasks on the human user's behalf, such as making and publishing marketing content, running accounts, organizing email and scheduling, etc.

MCP vs ACP: The Two Protocols Every AI Builder Needs to Know

Most people building with AI already know about MCP, but most people don't know about a different protocol that works alongside MCP: ACP.

In this video I break down everything you need to know about ACP whether you're technical or non-technical, how ACP works and how you can build applications using ACP today.

OpenAI launches GPT-5.5, calling it “a new class of intelligence”

On Thursday, OpenAI released GPT-5.5 and GPT-5.5 Pro. The new model is, unsurprisingly, its most capable model yet.

After Anthropic launched Opus 4.7 a week ago, it was only a matter of time before OpenAI would follow suit — and, at least according to the benchmarks we’ve seen so far, GPT-5.5 and 5.5 Pro beat Opus 4.7 across many of today’s standard benchmarks. In many benchmarks, though, GPT-5.4 Pro still outperforms the default GPT-5.5.

Kimi K2.6: BEST Opensource AI Model That Beats Opus 4.6 and Gemini 3.1 Pro

In this video, I fully test Kimi K2.6 across real-world tasks, from long-horizon coding to agent workflows and even front-end website generation. This model isn’t just competitive — it’s going head-to-head with top proprietary models like Opus 4.6 and Gemini 3.1 Pro… and in some cases, outperforming them.

Google’s Gemini can now run on a single air-gapped server —
and vanish when you pull the plug

Cirrascale Cloud Services today announced it has expanded its partnership with Google Cloud to deliver the Gemini model on premises through Google Distributed Cloud, making it the first neocloud provider to offer Google's most advanced AI model as a fully private, disconnected appliance.

Anthropic Built It. OpenAI and LangChain Just Responded. You Now Have A Decision To Make.

Three of the biggest players in AI all made moves in the agent space over the last seven days. Anthropic launched Claude Managed Agents, a fully cloud-hosted agent platform. The very next day, LangChain responded with Deep Agents Deploy, explicitly positioned as the open alternative. And then OpenAI shipped the next evolution of the Agents SDK.

The problem is none of these announcements make sense in isolation. Each one is solving a different version of the same problem, at a different point on the build-to-buy spectrum.

Mozilla launches Thunderbolt AI client with focus on
self-hosted infrastructure

The newly announced Thunderbolt is being sold as a front-end client for users and businesses who want to run their own self-hosted AI infrastructure without relying on cloud-based third-party services.

Thunderbolt is built on top of Haystack, an existing open source AI framework that lets users build custom, modular AI pipelines from user-chosen components. Thunderbolt acts as what Mozilla calls a “sovereign AI client” on top of that underlying infrastructure.