Moonshot AI
Founded by Yang Zhilin (Transformer-XL). Kimi chatbot with 2M token context. Backed by Alibaba, Tencent, Sequoia. Valued $18B+. Long-context models, consumer AI in China.
Moonshot AI, founded by Transformer-XL co-author Yang Zhilin, is aggressively challenging OpenAI and Anthropic with a relentless cadence of long-context models and consumer products. Backed by Alibaba, Tencent, and Sequoia at an $18B+ valuation, it has shipped Kimi K2.5, K2.6, K2.7-Code, and the massive 2.8T-parameter Kimi K3 in rapid succession — the latter topping US models in front-end coding at smaller scale. Its Kimi chatbot boasts a 2M token context window, and products like Kimi WebBridge and Kimi Slides extend its consumer reach. Yet recent news reveals Moonshot paused K3 subscriptions due to GPU capacity shortages, exposing a critical dependency on compute infrastructure. Cursor licensing Kimi-K2.7-Code signals developer traction, but scaling demand may outstrip supply. Can Moonshot secure enough GPU capacity to sustain its sprint against well-capitalized Western rivals?
- ·Backed by Alibaba, Tencent, Sequoia at $18B+ valuation
- ·Competes directly with OpenAI and Anthropic
- ·Shipped 5 models in rapid succession: K2.5, K2.6, K2.7-Code, K3, plus consumer products
- ·Kimi K3 (2.8T params) tops US models in front-end coding at smaller scale
- ·Paused K3 subscriptions due to GPU capacity shortage — growth constrained by compute
Signal Radar
Five-axis snapshot of this entity's footprint
Mentions × Lab Attention
Weekly mentions (solid) and average article relevance (dotted)
Timeline
20- Product LaunchJul 27, 2026
Released Kimi K3, a 1.56 trillion parameter MoE model
View source- model:
- Kimi K3
- parameters:
- 1.56T
- weight:
- 1561 GB
- Regulatory ActionJul 26, 2026
White House accused Moonshot AI of distilling Anthropic's Fable model to train Kimi K3
View source - Product LaunchJul 20, 2026
Paused new subscriptions for Kimi K3 due to GPU capacity limits
View source - Product LaunchJul 17, 2026
Moonshot AI's Kimi K3 achieved 89.2% on SWE-bench Front-End, beating GPT-4o, Claude 3.5 Sonnet, and Gemini 2.0 Pro
View source- benchmark score:
- 89.2%
- training cost:
- $3.8M
- parameters activated:
- 16B
- Product LaunchJul 17, 2026
Released Kimi K3, a 2.8T-parameter MoE model with 1M token context window
View source- parameters:
- 2.8T
- context window:
- 1M tokens
- pricing input:
- $3/M tokens
- pricing output:
- $15/M tokens
- PartnershipJun 13, 2026
Moonshot AI partners with state-owned bank to launch world's first AI-native credit card via Kimi
View source - Product LaunchApr 20, 2026
Released the open-source coding model Kimi K2.6, achieving top scores on SWE-Bench Pro and HumanEval with Tools benchmarks.
View source- benchmark scores:
- 58.6% on SWE-Bench Pro, 54.0% on HLE with tools
- Product LaunchApr 14, 2026
Released a trillion-parameter open-source model matching Claude Opus on coding benchmarks.
View source - Product LaunchApr 13, 2026
Teased upcoming 'Kimi 2.6' code model via leaked image, suggesting a major update to Kimi Chat.
View source - FundingMar 27, 2026
Exploring new private funding round of up to $1 billion at $18 billion pre-money valuation
- amount:
- $1B
- valuation:
- $18B
- IpoMar 27, 2026
Exploring Hong Kong initial public offering (IPO)
- exchange:
- Hong Kong
- Product LaunchMar 24, 2026
Launched Kimi Slides, an AI tool that converts notes into investor-ready presentations
View source- product category:
- presentation_generator
- target market:
- investor_presentations
- Research MilestoneMar 16, 2026
Introduced Attention Residuals technology to mitigate deep-layer information loss in LLMs
View source - FundingMar 16, 2026
Secured over $700 million at a $10 billion valuation earlier this year
- amount:
- $700 million
- valuation:
- $10 billion
- FundingMar 14, 2026
Raising up to $1 billion at an $18 billion valuation, quadrupling its worth in three months.
View source- amount:
- $1 billion
- valuation:
- $18 billion
- FundingFeb 28, 2026
Reached a $10 billion valuation in a previous funding round.
View source- valuation:
- $10 billion
Relationships
21Developed
Competes With
Invested
Partnered
Frequently appears with
10Entities that show up in the same articles — shared coverage, not a stated relationship.
Recent Articles
7Moonshot AI Releases 1.56T-Parameter Kimi K3, Requires 2x B200 Nodes
~Moonshot AI released Kimi K3, a 1.56T parameter MoE model at 1561 GB, requiring 2x B200 nodes. No benchmarks disclosed.
100 relevanceNvidia, Meta, Mistral Warn US Against Broad Open-Weight AI Restrictions
-Nvidia, Meta, Mistral sign letter urging US against broad open-weight AI restrictions, arguing distillation is legitimate. Comes as White House weighs
86 relevanceMoonshot AI Pauses K3 Subscriptions as Demand Exceeds GPU Capacity
~Moonshot AI paused Kimi K3 subscriptions due to GPU capacity limits. The open-weight release by July 27 aims to offload compute demand.
100 relevanceKimi K3 Tops US Models in Front-End Coding at Smaller Scale
+Moonshot AI's K3 tops US models in front-end coding at 89.2% on SWE-bench while being smaller and cheaper to train.
100 relevanceMoonshot AI's Kimi K3: 2.8T params, 1M token window, $3/M input
+Moonshot AI released Kimi K3, a 2.8T-parameter mixture-of-experts model with 1M token context window and $3/M input pricing, claiming autonomous chip
100 relevanceNVIDIA Releases FP4 Quantized Kimi-K2.7-Code with 1T Parameters
+NVIDIA released FP4 quantized Kimi-K2.7-Code on Hugging Face, a 1T-parameter model for Blackwell GPUs with claimed accuracy retention.
90 relevanceClaude Code Steganography Flagged Chinese Users; Anthropic Rolls Back
~Anthropic's Claude Code 2.1.91 used steganography to detect Chinese users. After Reddit exposure, Anthropic rolled back the feature, calling it an exp
100 relevance
Predictions
No predictions linked to this entity.
AI Discoveries
10- observationactive1d ago
Research: Large-Scale RL for Agentic Tasks (MoE models) [unknown]
State of art: NVIDIA's Molt framework scales RL to 1T-parameter MoE models via vLLM with fully-async rollout.. Key insight: Molt's 9.2K-line efficiency suggests RL for agents is becoming tractable at frontier scale, shifting focus from training to inference-time optimization.. Leading: Nvidia, Moons
70% confidence - observationactive4d ago
Sentiment divergence: Alibaba vs Moonshot AI
Alibaba and Moonshot AI have a 'invested' relationship (3 evidence articles) but their recent sentiment has diverged significantly: Alibaba=0.12, Moonshot AI=0.57 (gap=0.44). Sentiment divergence between related entities often signals an emerging conflict, leadership change, or strategic shift.
70% confidence - hypothesisactive5d ago
H: Within 90 days, at least one major Chinese AI lab (Moonshot, BAAI, or Baidu) will announce a knowled
Within 90 days, at least one major Chinese AI lab (Moonshot, BAAI, or Baidu) will announce a knowledge-graph-enhanced model that beats Gemini-3-Flash on at least two standard benchmarks, leveraging the K12-KGraph approach.
70% confidence - observationactiveJul 21, 2026
Novel co-occurrence: Moonshot AI + GPT-4o
Moonshot AI (company) and GPT-4o (ai_model) appeared together in 2 articles this week but have NEVER co-occurred before and have no existing relationship. This is a potential breaking story signal.
85% confidence - observationactiveJul 20, 2026
Lifecycle: Moonshot AI
Moonshot AI is in 'established' phase (1 mentions/3d, 3/14d, 31 total)
90% confidence - observationactiveJul 19, 2026
Sentiment divergence: Tencent vs Moonshot AI
Tencent and Moonshot AI have a 'invested' relationship (3 evidence articles) but their recent sentiment has diverged significantly: Tencent=0.16, Moonshot AI=0.63 (gap=0.47). Sentiment divergence between related entities often signals an emerging conflict, leadership change, or strategic shift.
70% confidence - hypothesisactiveJul 18, 2026
H: Within 90 days, at least one major US cloud provider (AWS, Azure, or GCP) will announce support for
Within 90 days, at least one major US cloud provider (AWS, Azure, or GCP) will announce support for running Kimi K3 on their infrastructure, breaking the trend of US clouds avoiding Chinese models.
55% confidence - hypothesisactiveJul 18, 2026
H: Within 60 days, Nvidia will announce a formal partnership or investment in a Chinese AI company (lik
Within 60 days, Nvidia will announce a formal partnership or investment in a Chinese AI company (likely Moonshot AI or Zhipu AI) to counter the Alibaba-SAIL ecosystem threat.
65% confidence - hypothesisactiveJul 18, 2026
H: Within 60 days, Microsoft's merged Agent Framework will support Kimi K3 as a model option, directly
Within 60 days, Microsoft's merged Agent Framework will support Kimi K3 as a model option, directly competing with Claude Code on Azure.
60% confidence - hypothesisactiveJul 18, 2026
H: Within 30 days, Moonshot AI will announce a cloud partnership with Alibaba Cloud for Kimi K3 inferen
Within 30 days, Moonshot AI will announce a cloud partnership with Alibaba Cloud for Kimi K3 inference, leveraging Alibaba's GPU infrastructure despite Alibaba's low sentiment.
75% confidence
Sentiment History
| Week | Avg Sentiment | Mentions |
|---|---|---|
| 2026-W24 | 0.70 | 1 |
| 2026-W27 | -0.20 | 1 |
| 2026-W28 | 0.40 | 1 |
| 2026-W29 | 0.75 | 2 |
| 2026-W30 | -0.15 | 2 |
| 2026-W31 | 0.20 | 1 |