funding_articleMay 20, 2026Confidence: high
Cactus | Y Combinator
Summer 2025 batch; low-latency on-device AI engine for mobile and wearables.
Affected score: yesObserved: May 20, 2026Round: Pre-Seed
raisedpre-seed
C
Cactus
Pre-SeedMarket data is refreshed once per day from public sources. Information may be incomplete or outdated — verify independently before making decisions. This is not investment advice.
Beta
DealFlow OS uses public web data and automated enrichment. Research may be incomplete, outdated, or incorrect. Verify important information before making investment or outreach decisions.
Investor read
Evidence-bound summary — expand sections for movement, risks, and signals.
Memo snapshot · May 20, 2026, 6:21 PM
Beta
DealFlow OS uses public web data and automated enrichment. Research may be incomplete, outdated, or incorrect. Verify important information before making investment or outreach decisions.
What they do
Cactus - On-device AI for Smartphones, Laptops & Edge One inference engine for on-device AI across smartphones, laptops, and edge hardware
Funding
Raised $500K across 1 funding round. Latest: $500K Pre-seed (Jun 2025). Investors: Y Combinator. (High).
Quick read
- •Cactus - On-device AI for Smartphones, Laptops & Edge One inference engine for on-device AI across smartphones, laptops, and edge hardware
- •Reported angle: TurboQuant-H: Hadamard Rotation for 2-Bit Embedding Quantization
- •Indexed activity snapshot: 1 funding‑related row(s), 0 hiring‑related, 0 GitHub‑tagged, 7 product/news‑style — scoring reflects corpus coverage only.
Stage
Seed (YC)
Evidence summary
Verified facts
- •Cactus - On-device AI for Smartphones, Laptops & Edge One inference engine for on-device AI across smartphones, laptops, and edge hardware
- •Reported angle: TurboQuant-H: Hadamard Rotation for 2-Bit Embedding Quantization
- •Indexed activity snapshot: 1 funding‑related row(s), 0 hiring‑related, 0 GitHub‑tagged, 7 product/news‑style — scoring reflects corpus coverage only.
Recent movers
- •May 11, 2026 · Blog / news — TurboQuant-H: Hadamard Rotation for 2-Bit Embedding Quantization
- •May 11, 2026 · Blog / news — Ridiculously Fast On-Device Transcription: Reviewing Parakeet CTC 1.1B with Cactus
- •May 11, 2026 · Blog / news — LFM-2.5-350m on Cactus: 140 tok/sec, Single Core, 355 MB
+5 more in Recent movement below
- •May 11, 2026 · Blog / news — TurboQuant-H: Hadamard Rotation for 2-Bit Embedding Quantization
- •May 11, 2026 · Blog / news — Ridiculously Fast On-Device Transcription: Reviewing Parakeet CTC 1.1B with Cactus
- •May 11, 2026 · Blog / news — LFM-2.5-350m on Cactus: 140 tok/sec, Single Core, 355 MB
- •May 11, 2026 · Blog / news — The Sweet Spot for Mac Code Use: Reviewing LFM2 24B MoE A2B with Cactus
- •May 11, 2026 · Blog / news — Sub-150ms Transcription with Cloud-Level Accuracy: Why We Built a Hybrid Engine
- •May 11, 2026 · Blog / news — Gemma 4 on Cactus: The first model you can talk to, show things, and trust to know when it needs help
- •May 11, 2026 · Blog / news — Engineering Blog | Cactus
- •May 20, 2026 · ycombinator.com — Cactus | Y Combinator
- •Y Combinator
Funding
Raised $500K across 1 funding round. Latest: $500K Pre-seed (Jun 2025). Investors: Y Combinator. (High).
Hiring
No hiring/careers evidence indexed (Low).
GitHub
No GitHub‑linked evidence indexed (Low).
Product / news
7 product/news‑styled row(s); headline risk without filings (High).
Traffic / social
No traffic/social evidence indexed (Low).
Funding & hiring signals
Open roles (indexed)
No open roles indexed yet.
Failed or blocked links
- public_page:_pressnot_foundLast checked Mon, May 11, 04:56 AM
HTTP 404
- public_page:_newsnot_foundLast checked Mon, May 11, 04:56 AM
HTTP 404
- public_page:_jobsnot_foundLast checked Mon, May 11, 04:56 AM
HTTP 404
- public_page:_companynot_foundLast checked Mon, May 11, 04:56 AM
HTTP 404
- public_page:_careersnot_foundLast checked Mon, May 11, 04:56 AM
HTTP 404
- public_page:_aboutnot_foundLast checked Mon, May 11, 04:56 AM
HTTP 404
DealFlow growth score
42.0Limited recent public signal
7D+0%
30D+0%
Needs Review
The score is an algorithmic estimate based on observed public company-level signals. It may be incomplete, stale, or inaccurate and is not investment, legal, tax, or business advice.
Source health
- public_market_enrichmentokLast checked Mon, May 11, 04:56 AM
- public_page:_blog_turboquant-hokLast checked Mon, May 11, 04:56 AM
- public_page:_blog_parakeetokLast checked Mon, May 11, 04:56 AM
- public_page:_blog_lfm2-5-350mokLast checked Mon, May 11, 04:56 AM
- public_page:_blog_lfm2-24b-a2bokLast checked Mon, May 11, 04:56 AM
- public_page:_blog_hybrid-transcriptionokLast checked Mon, May 11, 04:56 AM
- public_page:_blog_gemma4okLast checked Mon, May 11, 04:56 AM
- public_page:_blogokLast checked Mon, May 11, 04:56 AM
- public_page:_okLast checked Mon, May 11, 04:56 AM
- public_page:homeokLast checked Mon, May 11, 04:56 AM
DealFlow score momentum
427D +030D +0
100500
More runs will build history.
The score is an algorithmic estimate based on observed public company-level signals. It may be incomplete, stale, or inaccurate and is not investment, legal, tax, or business advice.
Signal breakdown
Latest momentum signal per category. Expand a card to inspect raw payloads.
Public source summary
- Total evidence rows
- 11
- Latest evidence
- Wed, May 20, 06:21 PM
Source types found
blogcompany_sitefunding_articleofficial_sitepress
Strongest / recent news-style rows
Cactus | Y Combinator
Wed, May 20, 06:21 PM · confidence 88%high quality
https://www.ycombinator.com/companies/cactus
Public signal timeline
Newest first · 11 event(s)
1
Wed, May 20, 06:21 PM · funding_article · 88% · verified_publichigh quality
Source ↗Cactus | Y Combinator
Summer 2025 batch; low-latency on-device AI engine for mobile and wearables.
2
Wed, May 20, 06:21 PM · company_site · 85% · verified_publichigh quality
Source ↗Cactus - On-device AI for Smartphones, Laptops & Edge
One inference engine for on-device AI across hardware targets.
3
Mon, May 11, 04:56 AM · blog · 90% · publichigh quality
Source ↗TurboQuant-H: Hadamard Rotation for 2-Bit Embedding Quantization
Source: Blog / news
A simplified offline variant of TurboQuant using Hadamard rotation and per-group Lloyd-Max codebooks — 4× compression of per-layer embeddings in Gemma 4 E2B at +0.06 PPL.
4
Mon, May 11, 04:56 AM · blog · 90% · publichigh quality
Source ↗Ridiculously Fast On-Device Transcription: Reviewing Parakeet CTC 1.1B with Cactus
Source: Blog / news
Review of NVIDIA's Parakeet-CTC-1.1B model running locally on Mac with Cactus. Architecture breakdown, benchmarks, and transcription use cases.
5
Mon, May 11, 04:56 AM · blog · 90% · publichigh quality
Source ↗LFM-2.5-350m on Cactus: 140 tok/sec, Single Core, 355 MB
Source: Blog / news
Benchmarking Liquid's LFM-2.5-350m across seven devices with Cactus. INT8 quantization, single-core CPU decode, zero-copy loading, and why this configuration makes on-device inference practical.
6
Mon, May 11, 04:56 AM · blog · 90% · publichigh quality
Source ↗The Sweet Spot for Mac Code Use: Reviewing LFM2 24B MoE A2B with Cactus
Source: Blog / news
Review of LiquidAI's LFM2-24B-A2B mixture-of-experts model running locally on Mac with Cactus. Architecture breakdown, benchmarks, and coding agent use cases.
7
Mon, May 11, 04:56 AM · blog · 90% · publichigh quality
Source ↗Sub-150ms Transcription with Cloud-Level Accuracy: Why We Built a Hybrid Engine
Source: Blog / news
How Cactus combines on-device and cloud inference for real-time speech transcription with sub-150ms latency and automatic cloud handoff for noisy audio.
8
Mon, May 11, 04:56 AM · blog · 90% · publichigh quality
Source ↗Gemma 4 on Cactus: The first model you can talk to, show things, and trust to know when it needs help
Source: Blog / news
Gemma 4 runs natively on your device with real-time voice, vision, and audio, and routes hard problems to the cloud when it should.
9
Mon, May 11, 04:56 AM · blog · 90% · publichigh quality
Source ↗Engineering Blog | Cactus
Source: Blog / news
Deep dives into on-device AI, inference optimization, and running models on smartphones, laptops, and edge hardware.
10
Mon, May 11, 04:56 AM · official_site · 90% · publichigh quality
Source ↗Cactus - On-device AI for Smartphones, Laptops & Edge
Source: Homepage
One inference engine for on-device AI across smartphones, laptops, and edge hardware. Run LLMs, transcription, and embeddings locally with automatic cloud fallback.
11
Wed, May 20, 06:21 PM · press · 85% · verified_publiclow quality
Source ↗cactus-compute/cactus on GitHub
Open-source low-latency mobile AI engine.
Official / company site
2 row(s)
company_site·Wed, May 20, 06:21 PM·Confidence 85%high qualityverified_public
Cactus - On-device AI for Smartphones, Laptops & Edge
One inference engine for on-device AI across hardware targets.
https://www.cactuscompute.comofficial_site·Mon, May 11, 04:56 AM·Confidence 90%high qualitypublic
Cactus - On-device AI for Smartphones, Laptops & Edge
Source name: Homepage
One inference engine for on-device AI across smartphones, laptops, and edge hardware. Run LLMs, transcription, and embeddings locally with automatic cloud fallback.
https://cactuscompute.com/Funding / news
1 row(s)
funding_article·Wed, May 20, 06:21 PM·Confidence 88%high qualityverified_public
Cactus | Y Combinator
Summer 2025 batch; low-latency on-device AI engine for mobile and wearables.
https://www.ycombinator.com/companies/cactusGitHub
1 row(s)
press·Wed, May 20, 06:21 PM·Confidence 85%low qualityverified_public
cactus-compute/cactus on GitHub
Open-source low-latency mobile AI engine.
https://github.com/cactus-compute/cactusBlog
7 row(s)
blog·Mon, May 11, 04:56 AM·Confidence 90%high qualitypublic
TurboQuant-H: Hadamard Rotation for 2-Bit Embedding Quantization
Source name: Blog / news
A simplified offline variant of TurboQuant using Hadamard rotation and per-group Lloyd-Max codebooks — 4× compression of per-layer embeddings in Gemma 4 E2B at +0.06 PPL.
https://cactuscompute.com/blog/turboquant-hblog·Mon, May 11, 04:56 AM·Confidence 90%high qualitypublic
Ridiculously Fast On-Device Transcription: Reviewing Parakeet CTC 1.1B with Cactus
Source name: Blog / news
Review of NVIDIA's Parakeet-CTC-1.1B model running locally on Mac with Cactus. Architecture breakdown, benchmarks, and transcription use cases.
https://cactuscompute.com/blog/parakeetblog·Mon, May 11, 04:56 AM·Confidence 90%high qualitypublic
LFM-2.5-350m on Cactus: 140 tok/sec, Single Core, 355 MB
Source name: Blog / news
Benchmarking Liquid's LFM-2.5-350m across seven devices with Cactus. INT8 quantization, single-core CPU decode, zero-copy loading, and why this configuration makes on-device inference practical.
https://cactuscompute.com/blog/lfm2-5-350mblog·Mon, May 11, 04:56 AM·Confidence 90%high qualitypublic
The Sweet Spot for Mac Code Use: Reviewing LFM2 24B MoE A2B with Cactus
Source name: Blog / news
Review of LiquidAI's LFM2-24B-A2B mixture-of-experts model running locally on Mac with Cactus. Architecture breakdown, benchmarks, and coding agent use cases.
https://cactuscompute.com/blog/lfm2-24b-a2bblog·Mon, May 11, 04:56 AM·Confidence 90%high qualitypublic
Sub-150ms Transcription with Cloud-Level Accuracy: Why We Built a Hybrid Engine
Source name: Blog / news
How Cactus combines on-device and cloud inference for real-time speech transcription with sub-150ms latency and automatic cloud handoff for noisy audio.
https://cactuscompute.com/blog/hybrid-transcriptionblog·Mon, May 11, 04:56 AM·Confidence 90%high qualitypublic
Gemma 4 on Cactus: The first model you can talk to, show things, and trust to know when it needs help
Source name: Blog / news
Gemma 4 runs natively on your device with real-time voice, vision, and audio, and routes hard problems to the cloud when it should.
https://cactuscompute.com/blog/gemma4blog·Mon, May 11, 04:56 AM·Confidence 90%high qualitypublic
Engineering Blog | Cactus
Source name: Blog / news
Deep dives into on-device AI, inference optimization, and running models on smartphones, laptops, and edge hardware.
https://cactuscompute.com/blogPrivate workspace
Sign in as an active team member to view private notes, watchlist controls, transcript evidence, and interaction history.