Privacy and advertising choices

Git-Stars uses essential storage for site operation. Optional analytics and ad-measurement scripts stay disabled unless you accept them; partners such as Google may then use cookies or similar identifiers where required. Privacy Policy

LogoGit-Stars
Top StarsTrendingAI AgentsDaily PicksViral ReposInsights
LogoGit-Stars

Discover top GitHub projects with real rankings and AI insights

GitHub
Built withLogo of Git-StarsGit-Stars
Rankings
  • Top Stars
  • Trending
  • AI Agents
  • Daily Picks
  • Explore
Resources
  • Insights
  • Editorial Policy
About
  • About
  • Contact
Legal
  • Privacy Policy
  • Terms of Service
© 2026 Git-Stars. All Rights Reserved.
FK

MoonshotAI/FlashKDA

FlashKDA: high-performance Kimi Delta Attention kernels

981 Stars95 Forks18 Open Issues981 WatchersCudaMIT
Developer ToolUI Framework
Review Readiness

This repository page is useful for visitors, but Git-Stars keeps it out of search indexing until more original evidence and comparison context are available.

60

review score

Needs original analysisNeeds alternatives
Decision Snapshot

Problem solved

FlashKDA solves the need for high-performance attention kernels for Kimi Delta Attention, providing faster execution than Triton-based alternatives on modern NVIDIA GPUs.

Deployment reality

The available setup signal starts with: pip install -v --no-build-isolation .. Treat this as a starting point, then ask Codex or Claude to inspect the README, environment variables, runtime version, and deployment target before production use.

Commercial use

The recorded license is MIT. This is a useful commercial-use signal, but teams should still verify license text, dependencies, model/API terms, and trademark constraints.

Capability ceiling

MoonshotAI/FlashKDA still needs a clearer capability analysis. Use the metadata as a discovery signal, not as a production recommendation.

Source and compliance noteLast synced: Jul 30, 2026

Git-Stars is independent and not affiliated with GitHub or this project. Analysis may be AI-assisted and based on public repository metadata plus short README-derived summaries. We do not mirror full README files, docs, issues, or social comments.

Original GitHub sourceMethodologyEditorial Policy
Editorial Evaluation

MoonshotAI/FlashKDA is tracked as a Cuda project in the Developer Tool, UI Framework area. This evaluation combines public GitHub metadata, category signals, short source summaries, and Git-Stars editorial rules rather than copying project documentation.

Momentum check: the repository has 981 total stars, with +91 today, +0 this week, and +0 this month. These growth windows help distinguish durable adoption from short-lived visibility spikes.

Maintenance check: current activity is Active; the latest push was 26 days ago, and the open issue queue is 18, about 1.83% of total stars. Treat this as an adoption signal, not a substitute for engineering due diligence.

Adoption check: 95 forks and 981 watchers suggest how often the project is reused or followed. License signal: MIT. Always verify license compatibility before commercial or internal use.

Practical fit: this project is most relevant when you need Cuda teams evaluating ecosystem-native tooling. Be cautious when you need a legal review, security audit, or production SLA.

Source check: Git-Stars currently has 2 explicit source reference(s) for this report and a recent growth signal of 91. Follow the original GitHub repository for final install, security, and release information.

Evidence Checklist

Analysis

Limited

Needs stronger original analysis before indexing.

Sources

Ready

Repository metadata and README/source references are attached.

License

Ready

MIT is recorded for review.

Maintenance

Ready

Recent activity is visible in repository metadata.

Alternatives

Limited

Needs more comparable projects for decision support.

Best For
  • Cuda teams evaluating ecosystem-native tooling
  • use cases where recent maintenance matters
Avoid When
  • you need a legal review, security audit, or production SLA
Adoption Signals

Momentum

981 Stars

Reuse

95 Forks

Attention

981 Watchers

Maintenance

active

License

MIT

Open issues

18

Overview

FlashKDA is a high-performance KDA kernel library built on CUTLASS, optimized for SM90+ GPUs. It provides efficient implementations for Kimi Delta Attention, with auto-dispatch support from flash-linear-attention's chunk_kda.

Key Features

- High-performance KDA kernels built on CUTLASS for SM90+ GPUs - Auto-dispatch from flash-linear-attention's chunk_kda with fallback to Triton - Supports variable-length batching, optional initial/final recurrent states, and configurable gate parameters

Tool Positioning

Developer Tool

Tools that improve coding, testing, build, and local workflow

UI Framework

Frontend frameworks, design systems, and interface libraries

Quick Start
pip install -v --no-build-isolation .
View on GitHub
Project Activity

65

Health Score

Active

Commit Activity

Apr 20, 2026

Created

Jul 29, 2026

Last push

Source Trail

GitHub repository metadata

metadata

GitHub README

readme_summary

Star History

+0

Today's growth

+0

7-day growth

+0

30-day growth

Jul 30, 2026Jul 30, 2026
Community Health
95

Forks

18

Open

981

Watchers

Owner
FK

MoonshotAI

GitHub profile
Topics & Language
Cuda
Ecosystem & Usage
GitHub Repository
Alternatives & Comparison

No similar projects found. Check Explore for more.

License
MIT
CreatedApr 20, 2026
Last pushJul 29, 2026
Last syncedJul 30, 2026
Repository Standards

✓

License

✓

Forked

✓ Active

Maintained