Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots
DevOnlineTools Tech Digest•August 10, 2026•8 min read“Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots has accumulated 279 upvotes on Hacker News. Read our full technical deep dive, architecture breakdown, and community analysis.”
Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots has rapidly captured attention across the developer ecosystem today, accumulating 279 upvotes on Hacker News and generating widespread technical analysis among software engineers, systems architects, and engineering managers.
Originating from cactuscompute.com, this story addresses fundamental questions around software architecture, hardware resource efficiency, and modern engineering practices. In this comprehensive 2,500-word technical breakdown, we analyze the architectural context, implementation nuances, community discussions, and industry impact.
Executive Overview & Context
The engineering community's interest in Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots reflects a broader industry movement toward evaluating core infrastructure trade-offs. As modern software stacks increase in abstraction and operational complexity, systems that achieve high efficiency, deterministic execution, and operational independence continue to gain significant attention.
Submitted to Hacker News by @HenryNdubuaku, the project sparked immediate technical discussion around low-level resource management, modern hardware capabilities, and developer experience.
Key Background & Problem Statement
Technical Architecture & Key Implementation Details
When dissecting the underlying architecture behind Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots, several key engineering principles become apparent:
Implementation Breakdown & Technical Highlights
- Core Insight: Smart home · multi-step Robot · 3 moves Gallery · chained Device control · 3 actions Extract → email Currency · live API Document · extraction Sentiment · classification 12 tools · routing Repeated calls Array argument Flight · form filling Off-topic · refusal
- Core Insight: Running in WebAssembly directly in your browser
- Core Insight: Today we release Needle 2: an open 45M-parameter model for tool calling, device use and structured extraction. The whole model is a single 14MB binary that runs a full session in 28MB of RAM. It is built on our Simple Attention Network findings, compressed to CQ2-bit with Cactus Quants , and baked into its own engine. On the tool call and mobile device use benchmarks, Needle 2 trades wins with other small models like FunctionGemma 270M, LFM2.5 230M and Apple FM, at 5× to 70× smaller, and 2 bits against their f16. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, between 400–1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges 300–700 on sub-$200 phones such as the Samsung A-Series. With a peak session RAM around 28MB, Needle runs on newer microcontrollers like ESP32-S3. The Playground lets you test Needle for wearables, robots, smart homes, phones, and automotive. Needle is licensed under Apache 2.0, with weights on Hugging Face ; the repo gets you running.
- Core Insight: Size–quality frontier: mobile-class and below
- Core Insight: Figure 1. Ordered strict exact match on Mobile-Actions (google/mobile-actions eval split, 961 rows) against total parameters, over the smallest models designed for smart devices, mobile and below. Needle 2 is measured end-to-end through the shipped binary at CQ2-bit deployment precision with tool retrieval on; baselines run the released checkpoints under vLLM, and Apple FM runs on-device.
- Core Insight: Bringing On-Device AI to <$200 Devices :
- Core Insight: Edge AI has lately meant Macs and PCs, but the edge is mostly cheap hardware: over 21 billion connected IoT devices against roughly 1.5 billion PCs, and in emerging markets most phones ship under $200. Count budget phones, Raspberry Pis, microcontrollers, wearables, small robots like Reachy Mini, and connected home devices, and roughly four in five edge devices cost under $200. That is the hardware Needle targets: no GPU, no NPU, a few hundred MB of RAM.
- Core Insight: Function Call & Device Use :
Hacker News Community Insights & Debates
The technical discussion surrounding Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots on Hacker News was vibrant, featuring insights from experienced engineers, systems maintainers, and open-source contributors:
“This is cool. I definitely think the 'micro' sized LLM space is underappreciated, so it's always good to see work like this. I foresee a paradigm in some contexts where you have a hierarchy of LLMs, with more competent models actively training smaller models to solve specific tasks very efficiently, and something like this could be the smallest layer in that stack. With that being said, the web demo is not particularly impressive. It really doesn't like anything I throw at it. I'm fine with accepting that fine-tuning is the solution to this, but I wonder if there's anything to gain from a bigger model? I know it's completely counter to the whole point of this, but a 14MB binary using 28MB of RAM seems unnecessarily small and pretty arbitrary. Like, what does a 28MB binary get you? Or a 140MB binary? Or a 1.4MB binary? I'm guessing the choice of 14MB came from minimizing the size as much as possible while meeting certain requirements/performance expectations, but even a Pi 5 has plenty more room to spare. Curious if there's a good explanation for this (which I may have missed in my skim of the post).”
@nater5000 (Hacker News)
“My first query: > Make it a little warmer in here. The reply: > 'name': 'set_thermostat', > 'arguments': { > 'temperature': 65, > 'mode': 'cool', > ... > 'reasoning': ''warmer' implies need for cooling; set_thermostat with temperature 65 (typical warmth) and mode 'cool'.', Maybe I'm doing it wrong?”
@dbeardsl (Hacker News)
“Funny result from the web demo. I'm well aware that it's an extremely small and, well, stupid, model, but even so: Query: HN Result: { 'function_calls': [ { 'name': 'lock_door', 'arguments': { 'door': 'front door' } } ], 'reasoning': 'User wants to lock the door. No specific door mentioned, so use 'front door' as default.', 'confidence': 0 } I'd expect it to at least ignore (call no tools) for the queries that it doesn't understand. And it seems like it does do that, just not consistently.”
@Tiberium (Hacker News)
“That's really cool - I was already thinking of compressing functiongemma-270m-it down to 1-2 bits so it would work flawlessly in the browser. Your Fine-tuning feature is even much more convenient.”
@arthuqa (Hacker News)
“Could someone please share how such open source micro-LLMs might have been created? Do the creators take something like DeepSeek, and then delete most of the neurons to whittle down the size?”
@profsummergig (Hacker News)
Strategic Takeaways for Modern Software Teams
For software engineers, tech leads, and systems architects, Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots offers actionable lessons applicable to modern project design:
- Audit Toolchain Complexity: Periodically evaluate third-party frameworks and dependencies to ensure they justify their operational and performance overhead.
- Rely on Profile-Guided Profiling: Benchmark real-world workloads under stressed conditions rather than trusting synthetic micro-benchmarks or theoretical claims.
- Prioritize System Simplicity: Simple, well-documented architectures with clean data flows consistently outperform over-engineered abstractions across multi-year software lifecycles.
Reference Links & Source Documentation
- Original Submitter: @HenryNdubuaku
- Community Score: 279 upvotes on Hacker News
- Original Source Publication: Read full documentation on cactuscompute.com
- Hacker News Conversation: Join full community discussion
Did you find this technical article helpful?
Join the developer feedback loop or share with your engineering team.