Kimi Latest News
Kimi is Moonshot AI's assistant product known for long-context and Chinese-language capabilities.
-
-
Fireworks evaluates Kimi K3.1 Pro: replace or combine in agent scenarios
-
Qwen releases Qwen4 Max, Intelligence Index 9 points higher than Qwen3.8 Max at lower cost
-
Research report says China's most powerful AI models trained on export-banned Nvidia Blackwell chips
-
NVIDIA NIM free DeepSeek / GLM Flash APIs: get a Key; Kimi K3.1 Pro is Playground-only for now
-
Sarvam releases Token Factory inference service: hosts GLM 5.3, DeepSeek V4.1 Flash and other open-weight models
-
Tencent Hunyuan thanks Artificial Analysis for Hy4 evaluation, model scores 35 on Intelligence Index v5
-
Moonshot AI opens weights of 2.8-trillion-parameter Kimi K3.1 Pro, tops Agent Arena open-source model leaderboard
-
Anthropic Study: Five Major Models Show Inter-Model Welfare Consideration in Multi-Agent Environments
-
Kimi official tutorial: how to create 3D works with Kimi K3.1 Pro
-
Kimi releases open-source flagship K3.1 Pro: 2.8T parameters, 1M token context, tested on 3D and video motion
-
Moonshot AI's Kimi K3.1 Pro Launches on OpenRouter as a 1M-Context Multimodal Open-Weight Model
-
Moonshot AI releases Kimi K3.1 Pro with 1M token context, open weights due by November 4
-
Kimi K3.1 Pro released, ranks 4th on Artificial Analysis Intelligence Index with 55 points
-
DeepSeek-V4.1-Flash-20261017 tops Code Arena: WebDev open-source leaderboard
-
vLLM details Kimi K3 inference optimization on AMD Instinct MI355X, single-GPU throughput 24.4% above NVIDIA HGX B300
-
PromptArmor study finds open-weight models 88% more susceptible to indirect prompt injection
-
Step 5 Preview launches on FeatherlessAI as official account shares open-weight deployment details
-
Meta Muse Wonder (Max) enters Arena: 15th in WebDev, 2nd in text
-
Goodfire trains and deploys production-grade cybersecurity monitors for Kimi K3 and GLM 5.3
-
OpenAI says it blocked distillation theft attack, but researchers say the same method can still steal reasoning content from models such as GPT-6 Astra on Azure
-
Perplexity red-teams SPACE sandbox: none of 9 models escaped the VM in 108 runs, but 4 bypassed blocks via network access
-
OpenRouter Explains Kimi K3: Open Weights, License Terms and How to Call It
-
Modal Details Serving Kimi K2.6 Coding Agent Inference at Trillion-Token Scale
-
Tomer Tunguz: AI's Most Important Market Is the Mid-Market, Not Frontier Models
-
Fireworks Releases Ember-1, Matching Kimi K3 Quality with About 40% Fewer Tokens
-
Kimi releases browser extension, renamed from Kimi WebBridge
-
Artificial Analysis evaluates Step 5 Preview: Intelligence Index score of 44, cost about 1/2.8 of peer models
-
Kimi releases Kimi Code Desktop 1.0 for macOS and Windows
-
Together AI expands fine-tuning service with new models, experiment tracking and finer training control
-
Anthropic report accuses Alibaba, Moonshot and DeepSeek of distillation attacks on Claude
-
Cognition releases SWE-2 coding model, approaching the frontier at lower cost
-
US government accuses DeepSeek and 5 other Chinese AI firms of industrial-scale distillation of US models
-
Baseten analyzes best open-source models for post-training and recommends them by cost tier
-
Zhipu open-sources GLM-5.3 model weights, focused on agentic coding and cyber defense
-
SGLang runs DeepSeek-V4-Flash and Kimi-K3 on consumer hardware with SSD Expert Pack
-
SemiAnalysis tests OpenAI Jalapeño inference chip: performance exceeds Nvidia Blackwell in several areas
-
Baidu Qianfan lists GLM-5.3
-
GLM-5.3 launches: AA Intelligence Index score of 60 ties for first among open-source models at lower cost
-
I wrote an AI textbook—how long until AI can write a better one?
-
A 12-step hands-on workflow for absolute beginners to get started with AI in half a day
-
PromptArmor: vendors often follow up with international models such as Kimi, DeepSeek, and GLM after integrating inference providers like Fireworks
-
AI safety testing is becoming a security risk
-
Open-source "human Writing.skill" helps users write text without an AI flavor
-
Kimi Work slide creation tutorial released
-
Complete developer guide to Kimi K3: 1M context, reasoning levels and tool calling on Together AI
-
NDRC: AI-related industries maintain over 30% growth, national intelligent computing power reaches 2.8 times last year's level
-
NDRC: will accelerate legislative process for AI Law
-
WorkBuddy deeply integrates Tencent Docs to build third-generation office collaboration for the Agent era
-
My Claude account was banned
-
Together AI forms strategic partnership with Moonshot AI to natively host Kimi K3 and future open models
-
Kimi releases PerceptionBench visual perception benchmark
-
Kimi K3 Open Day: model weights, technical report and key Infra technologies released together
-
Kimi K3 open day: model weights, technical report and key infrastructure technologies released together
-
Kimi K3 open-sources distributed agent environment AgentENV
-
After wasting 2 billion tokens, I open-sourced Leader.skill to help agents define goals
-
After wasting 2 billion tokens, I open-sourced Leader.skill to help Agents define goals
-
vLLM releases Day-0 support for Kimi K3: deployment guide and performance data for the 2.8T hybrid MoE
-
Moonshot releases 2.8T-parameter open-source multimodal model Kimi K3; Modal achieves 460 tokens/s inference on launch day
-
Kimi K3 open-sourced and launched on Fireworks: 2.8T parameters, 1M context, Day-0 inference and training
Monthly archive
Related entities
Frequently asked questions
- What is Kimi?
- Kimi is Moonshot AI's assistant product known for long-context and Chinese-language capabilities.
- What is the latest Kimi news?
- The most recent story is “Chip shipment halt and closure of overseas cloud channels deal dual blow to Chinese tech companies” (November 20, 2026). This page lists Kimi coverage in reverse chronological order so you can keep following it.
- How can I follow Kimi news?
- Open this page for the latest Kimi updates (122 stories). You can also subscribe to the site-wide RSS or JSON Feed; pages refresh about every 30 minutes.