Chinese AI Startup Moonshot AI Launches New Open Model

Chinese AI startup Moonshot AI, backed by Alibaba, has unveiled a new open-source model, Kimi K2. It is claimed to excel in advanced knowledge, mathematics, programming, and agentic tasks.

🚀 Hello, Kimi K2! Open-Source Agentic Model!
🔹 1T total / 32B active MoE model
🔹 SOTA on SWE Bench Verified, Tau2 & AceBench among open models
🔹Strong in coding and agentic tasks
🐤 Multimodal & thought-mode not supported for now

With Kimi K2, advanced agentic intelligence… pic.twitter.com/PlRQNrg9JL

— Kimi.ai (@Kimi_Moonshot) July 11, 2025

In its announcement, the startup stated that its neural network surpassed Claude Opus 4 in two benchmarks and outperformed OpenAI’s GPT-4.1 in several metrics. Additionally, the model offers lower token processing costs:

• $0.15 per 1 million input tokens;
• $2.5 per 1 million output tokens.

In comparison, Claude Opus 4 charges $15 per 1 million input tokens and $75 per 1 million output tokens—100 and 30 times more, respectively. For OpenAI’s GPT-4.1, the figures are $2 and $8, respectively.

Kimi K2 is developed based on the Mixture-of-Experts architecture and includes 1 trillion parameters, of which 32 billion are active—representing specialized computational blocks used for specific tasks.

Kimi K2 is available for free through the Kimi app and web interface. The company has released the source code for two versions:

• Kimi-K2-Base—a basic model intended for researchers and developers who need full control over customization and tuning;
• Kimi-K2-Instruct—a post-trained version optimized for general use, including chatbots and agentic AI scenarios.

Initial feedback on Kimi K2 in English and Chinese social media is mostly positive, although some point out the common issue of hallucinations in the AI segment.

Kimi K2 is so good at tool calling and agentic loops, can call multiple tools in parallel and reliably, and knows “when to stop”, which is another important property.

It’s the first model I feel comfortable using in production since Claude 3.5 Sonnet. pic.twitter.com/TcEkPlBMuk

— Pietro Schirano (@skirano) July 13, 2025

“It’s the first model I feel comfortable using in production since Claude 3.5 Sonnet,” wrote MagicPath startup founder Pietro Schirano on X.

Open AI—A Growing Trend

The new model reflects a broader industry trend towards open-source code, enabling both startups and major tech companies to enhance efficiency and accelerate the deployment of their AI products. This approach involves providing public access to a program’s source code, allowing third-party developers to:

• modify and refine the model’s architecture;
• fix bugs;
• scale functionality;
• implement their own improvements and adaptations for specific tasks.

Among notable Chinese startups, open-source code is offered by DeepSeek and Tencent.

As for American competitors, Meta and partially Google have taken this path.

Meanwhile, OpenAI CEO Sam Altman announced the postponement of the launch of their own open AI model indefinitely, citing the need for additional safety tests.

we planned to launch our open-weight model next week.

we are delaying it; we need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take us.

while we trust the community will build great things with this model, once weights are…

— Sam Altman (@sama) July 12, 2025

Back in March 2024, Elon Musk announced the release of Grok’s source code.