The single-agent era has officially ended: if you can't beat one, then go up against 300.

Has the Agent finally bid farewell to “solo efforts” and entered a new stage of evolution?

Just this morning, the Dark Side of the Moon officially released and open-sourced the latest flagship model of the Kimi series — Kimi K2.6, less than three months after the previous version K2.5 was launched. After release, it quickly gained immense popularity, with the official Weibo post reaching over 4 million views.

Currently, Agents struggle to handle complex engineering projects effectively. While they excel at completing specific tasks independently, team collaboration still needs improvement. How to break through this limitation has become the core goal of Kimi K2.6.

The new version explores how to enhance the Agent’s team collaboration capabilities: further strengthening the Agent Swarm (Agent Cluster) feature introduced in K2.5, by adapting frameworks like OpenClaw to reinforce proactive work. The brand-new Claw Group (Claw Cluster) adds organizational collaboration capabilities. This systematic stacking of features constructs an AI system that more closely resembles a human team.

To achieve all this, the underlying model must be sufficiently powerful. Kimi K2.6 shows significant progress in core abilities such as general Agents, coding, and image understanding. Tests like Humanity’s Last Exam, SWE-Bench Pro (closer to real development scenarios), and DeepSearchQA (assessing deep retrieval capabilities) all show K2.6 firmly ahead of competitors.

Even when comparing K2.6 with closed-source models like GPT-5.4, Claude Opus 4.6, and Gemini 3.1 Pro, it holds its own, with some metrics even surpassing them.

The latest results from the AI evaluation platform Artificial Analysis declare, “Kimi K2.6 becomes the new king among open-source models”!

After launching Kimi K2.6, the large model aggregation platform OpenRouter gave high praise, noting that the Dark Side of the Moon’s new model emphasizes long-sequence programming capabilities, designed specifically for Agent scenarios requiring continuous execution. Compared to traditional chatbots, it functions more like a “systems engineer,” capable of breaking down complex tasks, executing step-by-step, and continuously optimizing during the process.

Some netizens have exclaimed that this flagship Kimi model is ridiculously powerful—already rivaling GPT-5.4 in coding tasks, at a much lower price than Opus 4.7, and it’s open-source and free to use. Now, every few months, a new open-source model approaches the performance of closed-source GPT and Claude. “It feels like open-source models are really catching up, and China is leading the pace.”

12 hours straight, 300 Agents working simultaneously

Is the ultimate form of Agent here?

This time, Kimi K2.6 continues to push forward in the coding domain. A few days ago, overseas discussions heated up over the low-profile launch of Kimi K2.6-Code-Preview, with high expectations for the official K2.6 release.

As the most powerful Kimi model for coding so far, K2.6 has achieved breakthroughs in long-term coding capabilities, helping to push software development automation into deeper engineering stages.

For example, Kimi K2.6 can successfully download Qwen3.5-0.8B locally on Mac and run it. It doesn’t follow common tech stacks but rewrites reasoning processes in the niche Zig language and continuously optimizes, demonstrating the model’s generalization ability.

The entire process lasted over 12 hours, during which it called tools more than 4,000 times, iterated 14 times. Through constant tuning and restructuring, inference speed jumped from about 15 tokens/sec initially to approximately 193 tokens/sec, ultimately being about 20% faster than local large-model chat applications like LM Studio.

The focus of this K2.6 upgrade is to further enhance the Agent cluster’s collaborative output ability. Simply put, it aims to clarify “how Agents work together.”

What can it do now? K2.6 can automatically decompose a complex task, assign different specialized Agents to handle search, in-depth research, document analysis, long-form writing, etc., then stitch the results together to continue progressing.

Under this mechanism, a single run can complete the entire chain: from raw data and web content to PPTs and spreadsheets—all automatically generated, without switching tools or manual handoffs.

Meanwhile, the underlying architecture of the Agent cluster has been expanded to support up to 300 sub-Agents working simultaneously, completing 4,000-step collaborations. Its parallel processing capability has been elevated to a new level. As the scale grows, the AI’s role shifts: it begins to take over the entire process and directly produce systematic results.

For example, the Agent cluster deconstructed and reused a dense astrophysics paper with visual data, generating about 7,000 words of research report, 20k data points, and 14 charts.

To evolve AI into a 24/7, uninterrupted cyber employee, Kimi K2.6 has deeply integrated frameworks like OpenClaw and Hermes Agent.

It further enhances the model’s autonomous execution: whether it’s the accuracy of API calls, stability during long runs, or safety protections during complex research tasks, K2.6 performs impressively.

In Vibe Coding, Kimi K2.6’s website design is more eye-catching. The generated site, especially the first screen, has a strong visual impact, with consistent style. The addition of interactive elements and scrolling effects also helps retain users longer.

Besides frontend design, K2.6 also brings surprises to backend developers: it now supports Kimi account login and form data collection. You can create event registration pages and easily view backend registration info. This makes front-end and back-end integration smoother.

Currently, Kimi K2.6 is the default model for Kimi’s web version, app, and Kimi Code programming assistant. Go ahead and try it out.

Hands-on testing, impressing everyone

Without further ado, let’s directly test some cases and see how they perform.

For the first part of testing, we selected “K2.6 Agent,” evaluating both practicality and aesthetics, to see if it can produce eye-catching front-end effects.

Does anyone like “Persona 5”?

It’s a highly recognizable artistic style—a visual assault wrapped in manga aesthetics. It challenges aesthetic norms with highly irregular design, embedding themes of “rebelling against social mediocrity” directly into pixels and lines. It perfectly blends 2D design with 3D space, integrating manga symbols and visual expression deeply.

If we opened a small P5-style bar, what would its homepage look like?

We found that during front-end webpage construction, Kimi K2.6’s intelligence performs thorough testing, even simulating clicks:

Additionally, we made a little Easter egg: referencing the opening video of “Persona 5 Royal,” Kimi K2.6 created a short animation without any material provided.

We continued to request different styles: “Design a visually impactful homepage for an e-commerce platform, with a top navigation bar including brand logo, search box, shopping cart, login/register buttons; a hero section showcasing main promotions, bestsellers, or seasonal discounts; below the hero, recommended products or categories; and at the bottom or a prominent area, some user reviews of selected products.”

A single generation produced a homepage with very high completeness. Though slightly imperfect, we believe small issues can be fixed in one iteration, and the overall quality remains excellent.

Next, we tested the Agent cluster’s ability by creating a promotional brochure for Stanford University’s “2026 AI Index Report,” requiring it to deliver a webpage, tables, and PPT slides, all without providing any additional info or documents, testing the cluster’s collaborative writing performance.

We observed that each agent had its own badge, role description, and profile. When using the Agent cluster, you truly feel like a strategic leader mobilizing all resources, assigning tasks based on expertise, forming a working team that automatically executes your commands. It’s almost like having “trustworthy” written on their badges.

All the required content was ultimately produced: a shiny webpage, efficiently formatted PPT, and serious data tables.

Is the future of multi-agent collaboration here?

These series of tests demonstrate the powerful capabilities of Kimi K2.6 as the “base model” of the Agent era.

In the current “lobster craze” sparked by OpenClaw, the newly launched Claw Group points to a clear path for next-stage evolution of intelligent agents.

Currently, Claw Group is in limited internal testing.

This feature marks a new era of agent collaboration. You can connect various Agents running locally, on mobile, or in the cloud, each equipped with tools, skills, and memory, working together in a “group” to push tasks forward.

Here, K2.6 acts more like a dispatcher: assigning roles based on ability—who’s good at retrieval, who handles analysis, who produces content. If any part stalls, it can detect and reassign tasks or replace personnel, keeping the process moving.

Imagine preparing a complex report or developing a multi-layered project: the Claw Group’s Agents will discuss, coordinate, and adjust like a team of professionals, ultimately delivering a precise, comprehensive result.

This innovation not only breaks through traditional individual AI execution modes but also advances organizational intelligence. Its emergence makes “multiple AI agents working together” more of a reality.