Kimi K3
Kimi K3 is an advanced AI model developed by Moonshot AI, serving as a foundational architecture for specialized applications. Recent developments include Cognition releasing the SWE-2 coding model, which is trained specifically on Kimi K3. This integration highlights the model's utility in enhancing software engineering capabilities within the broader open-source AI landscape.
Kimi K3 is a multimodal model from Moonshot AI, first covered on 18 September 2026. It is available through an API. Its context window is 1M tokens.
Last updated
At a glance
- Made by
- Moonshot AI
- Type
- Multimodal model
- Availability
- API
- Licence
- Not stated
- Context window
- 1M tokens
- First seen
- 18 September 2026
Taken from the coverage, not from our own testing; "not stated" means none of the reports we read gave it. All tracked models →
10 stories about Kimi K3
-
Moonshot AI's Kimi K3 opens on Amazon Bedrock with 2.8 trillion parameters
Moonshot AI announced that its Kimi K3 model is now available on Amazon Bedrock. The model is described as the first open‑weight model to reach 2.8 trillion parameters and adds native vision support with a…
2 sources primary sourceUnite.AIAWS Machine Learning Blog -
US frontier AI firms warn of sophisticated distillation attacks; China says it may counter
Bloomberg reports that U.S. officials and frontier AI developers are increasingly alarmed that distillation attacks could let foreign actors replicate powerful Western models at a fraction of the cost. The concern is…
2 sourcesCNBC TechnologyTom's Hardware -
Moonshot says Kimi is used by Chinese investment banks and VCs, reportedly files Hong Kong IPO
Chinese AI startup Moonshot announced that its Kimi model is now being deployed for financial services, with integration to a suite of data providers such as S&P Global Market Intelligence, Crunchbase, Wind,…
1 sourceCNBC Technology -
China narrows US AI gap as US agencies accuse Chinese firms of model distillation
China is rapidly closing the artificial intelligence gap with the United States, challenging the long-standing American lead in model capabilities and research. Despite US restrictions on advanced chips, Chinese…
3 sourcesTom's HardwareArs Technica AIorlandosentinel.com -
Notion adds AI 'skills' library and custom agents to streamline team workflows
Notion announced version 3.7 of its platform, introducing a shared "skills" library that lets teams store reusable AI instructions. The library centralizes prompts—such as generating monthly business reviews or…
1 sourceNotion Releases -
Nvidia Rubin Ultra shifts to 8-hi HBM; 4-hi stacks emerge as inference cost optimum
SemiAnalysis reports a structural shift in AI hardware design, marking the end of the trend toward ever-increasing High Bandwidth Memory (HBM) density per chip. Nvidia’s upcoming Rubin Ultra accelerator will feature…
1 sourceSemiAnalysis -
Real-SWE benchmark tests AI on private enterprise codebases, revealing high failure rates
A new benchmark called Real-SWE has been released to evaluate frontier AI models on private, real-world enterprise codebases. Unlike synthetic or public datasets, this benchmark uses licensed tasks from actual…
1 source HN 46withspecific.com -
Comprehensive Reading List Maps Open-Source AI Model Landscape and US‑China Competition
The latest Interconnects post curates a growing body of analysis on open‑source AI models, covering strategic motivations, economic impact, and technical trends. It cites essays from industry leaders like Bill Gurley…
1 source HN 50Interconnects -
Cognition Releases SWE-2 Coding Model Trained on Kimi K3
Cognition has launched SWE-2, its most advanced coding model to date, post-trained using reinforcement learning on Moonshot AI's 2.8-trillion-parameter open model, Kimi K3. The model achieves a 50.0% score on the…
2 sources HN 61MarkTechPosttokenstead.ai -
TPUv7 Ironwood beats NVIDIA B200/B300 on inference cost by up to 50%
SemiAnalysis has released the first third-party benchmark results for Google’s TPUv7 Ironwood, marking a significant shift as the accelerator moves beyond internal use to compete directly with NVIDIA in external…
1 sourceSemiAnalysis
