Universe/Language/Text/Cactus Compute, Inc.
Venture candidateAutomated record

Cactus Compute, Inc.

On-device AI inference engine for smartphones, laptops, and edge hardware with automatic cloud fallback.

Visit website ↗
Rising signal56.7
Rising rank#24
Profile completeness57%
Last public activitySep 2, 2026
01

Usage, audience and developer pull

48.1KHF model downloads788.1 per model
417HF followers681 total likes
16.8KGitHub stars1.4K forks
61Model portfolio3 datasets · 0 Spaces

AI infrastructure picks-and-shovels

  • Shipping now
  • GitHub momentum
  • Multi-modality surface
02

What it publishes

Language/TextAudio/SpeechMultimodalRetrieval/Embeddings
Top tasks
text-generation (22), automatic-speech-recognition (10), image-text-to-text (7), feature-extraction (2), voice-activity-detection (2), speaker-recognition (1)
Model sizes
Unknown (57); 3-10B (2); <0.5B (2)
Base families
Gemma, Qwen
Libraries
cactus-needle, jax, mlx
03

How it ships

Original repositories16
Forked repositories4
Pushed in 90 days6
Archived share50%
Languages
Python, C++, TypeScript, Shell, Kotlin, Ruby
Topics
llm, llamacpp, llms, ai, llm-inference, on-device-ai, cactus, android, arm, edge
Latest release
cactus-compute/cactus · Aug 17, 2026
04

Latest observed moves

  1. GitHub pushRepository activity
  2. HF modelModel published or updated
  3. GitHub releaseRelease v2.1.0
  4. HF datasetDataset published or updated
05

Publicly evidenced tooling

Training
MLX
Compute
Not disclosed
Hardware
Not disclosed
Tracking
Not disclosed
Bottlenecks
Not disclosed

“Not disclosed” means no evidence was found in the sampled public metadata or top model card—not that the lab does not use the tool.

06

Nearest modality peers

Validate recurring revenue, workload retention, margins, and differentiation from clouds and open source.

  • HF member count is only a public-profile proxy
foundedYearfoundersOrLeadershipemployeeCountfundingbusinessModelproducts

Completeness: 8 of 14 tracked field groups. Unknowns are explicit so an analyst can close them.

Every claim should be auditable.

Platform counters are a point-in-time public snapshot. Editorial facts are only promoted after source review.

Know when the thesis changes.

Weekly deltas, funding and team alerts, portfolio watchlists, CSV exports and an analyst-ready API turn a static profile into a sourcing system.

See tracker plans →