Alternatives
Products that do what An LLM Running on a PS Vita does
Hello all, a couple of moons ago I ported karpathy's llama2.c code to run inference on the TinyStories 260K & 15M checkpoints on the on the PS Vita with the ability to download/delete the models on device. Runs showed that the 260K model ran at ~120 tok/s and at 15M ran at 1.8 tok/s, which probably could be a bit higher if it weren't a single threaded application. Had fun working on it as a weekend project, check it out for yourselves: https://github.com/callbacked/psvita-llm
- 1SU
Here's a project I've been working on for the last few months. It's a new (I think) algorithm, that allows to adjust smoothly - and in real time - how many calculations you'd like to do during inference of an LLM model. It seems that it's possible to do just 20-25% of weight multiplications instead of all of them, and still get good inference results. I implemented it to run on M1/M2/M3 GPU. The mmul approximation itself can be pushed to run 2x fast before the quality of output collapses. The inference speed is just a bit faster than Llama.cpp's, because the rest of implementation…
2024 · asciinema.org
- 2IV
The video demo runs a 7b Model on a normal gaming GPU. I think it already works quite well (accounting for the limited hardware power). :)
2024 · github.com
- 38F
Hi HN! I'm just sharing a project I've been working on during the LLM Efficiency Challenge - you can now finetune Llama with QLoRA 5x faster than Huggingface's original implementation on your own local GPU. Some highlights: 1. Manual autograd engine - hand derived backprop steps. 2. QLoRA / LoRA 80% faster, 50% less memory. 3. All kernels written in OpenAI's Triton language. 4. 0% loss in accuracy - no approximation methods - all exact. 5. No change of hardware necessary. Supports NVIDIA GPUs since 2018+. CUDA 7.5+. 6. Flash Attention support via Xformers. 7. Supports 4bit and 16bit…
2023 · github.com
- 4
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
27d ago · cactuscompute.com
- 5KR
I discovered that in LLM inference, keys and values in the KV cache have very different quantization sensitivities. Keys need higher precision than values to maintain quality. I patched llama.cpp to enable different bit-widths for keys vs. values on Apple Silicon. The results are surprising: - K8V4 (8-bit keys, 4-bit values): 59% memory reduction with only 0.86% perplexity loss - K4V8 (4-bit keys, 8-bit values): 59% memory reduction but 6.06% perplexity loss - The configurations use the same number of bits, but K8V4 is 7× better for quality This means you can run LLMs with 2-3× longer…
2025 · github.com
- 6TV
May 2026 · github.com
- 7WM
We wrote our inference engine on Rust, it is faster than llama cpp in all of the use cases. Your feedback is very welcomed. Written from scratch with idea that you can add support of any kernel and platform.
2025 · github.com
- 8IR
The Emotion Engine has 32 MB of RAM total, so the trick is streaming weights from CD-ROM one matrix at a time during the forward pass — only activations, KV cache and embeddings live in RAM. This means models bigger than the RAM can still run, they just read more from disc. Had to build a custom quantized format (PSNT), hack endianness, write a tokenizer pipeline, and most of the PS2 SDK from scratch (releasing that separately). The model itself is also custom — a 10M param Llama-style architecture I trained specifically for this. And it works. On real hardware.
Mar 2026 · github.com
- 9
- 10
- 11IR
Democratisation of local AI is key. I've been working on pushing the limits of commercial hardware, squeezing any extra bit possible. My Scientific Agentic AI hareness helped me to reallocate every single bit of it. I rewrote the Kernel, I went down the CUDA rabbit hole until I have been able to explain any bit and any ms of computational power involved in the process pushing the Qwen 30B-A3B from 8 tok7s to 19 tok/s with llama.cpp up to 22.2 tok/s with my project and 109 tok/s on not novel content and speeding up the prefill by 5-9X
Jul 2026 · github.com
- 12
- 13RA
Hi there, looking for feedback on my new project "Featherless.AI" The idea is to allow users to run all the models on hugging face instantly. Via the OpenAI API compatible endpoint. Why? Because its a real chore to download models and spin up GPUs, especially if you want to test multiple models. Not to mention GPUs cost multiple dollars an hour to rent. And if we want more people to use open source AI, we got to make it easier for them to try and play with all of them. So what if instead of spinning up dedicated GPUs per model (which is what every provider is doing) We can startup a LLM…
2024 · featherless.ai
- 14CI
One of the most frequent questions one faces while running LLMs locally is: I have xx RAM and yy GPU, Can I run zz LLM model ? I have vibe coded a simple application to help you with just that. Update: A lot of great feedback for me to improve the app. Thank you all.
2025 · can-i-run-this-llm-blue.vercel.app
- 15AO
I've built an airgapped Retrieval-Augmented Generation (RAG) system for question-answering on documents, running entirely offline with local inference. Using Llama 3, Mistral, and Gemini, this setup allows secure, private NLP on your own machine. Perfect for researchers, data scientists, and developers who need to process sensitive data without cloud dependencies. Built with Llama C++, LangChain, and Streamlit, it supports quantized models and provides a sleek UI for document processing. Check it out, contribute, or suggest new features!
2024 · github.com
- 16
I made this after seeing someone posit the idea online yesterday over lunch then spent some time refining it. So far it's pretty impressive IMO! Right now I am running Qwen3-30B-A3B on my 24gb unified memory m4 MacBook Pro at 50 tok/sec and this should definitely not be working for such a large model on my middling hardware. Things are detailed in the README to get up and running and DESIGN.md has details on all the choices and such made along the way.
23d ago · github.com
- 17KP
I thought it'd be interesting to use Linux PSI (Pressure Stall Information) for an LLM runtime to trim the KV cache. This is mainly useful imo for edge devices like the Jetson Orin super nano kit which have unified memory. I haven't benched much, but plan to do so more over time and see if I can make a real use of it as I run local LLMs. Let me know if it makes sense :P (I of course vibed this idea)
Jun 2026 · github.com
- 18LC
Warning: Inference takes ages. Pics & Media: https://x.com/VulcanIgnis/status/1893420241310335329
2025 · github.com
- 19LP
Hey HN, My team is open-sourcing the inference stack and fined-tuned models we use to create LLM-powered NPCs: https://github.com/GigaxGames/gigax The generative agents paper [1] pioneered the idea of prompting LLMs to create autonomous NPCs. But existing implementations require multiple calls to an LLM to make the agent plan its day, chat with people, and interact with its environment [2]. Our approach allows NPCs to be stepped at runtime with a single pass on consumer-grade hardware, with reasonable latency. To achieve this, we've fine-tuned open-source LLMs [3] to…
2024 · github.com
- 20IB
Built a simple web app that tells you which open-source LLMs will work on your hardware. It auto-detects your specs, shows compatible models from Hugging Face, gives realistic performance estimates (tokens/sec), and recommends quantization settings. You can also manually input specs to see "what if I upgraded my RAM?" Made this after wasting time downloading giant models only to find they crawled on my hardware. Hope it saves you some frustration!
2025 · caniusellm.com
- 21SY
Hey HN, If you tried running open-source models like Llama 3.1 70B or 405B, you might have noticed that it gets very expensive. The reason looks obvious enough that you might have stopped even before trying it! - GPUs are very expensive to buy or rent - Running the most performing LLMs need 4, 8 or even 16 top of the line Nvidia GPUs - And that won’t get you anywhere near the level of VRAM needed to batch enough to get a decent throughput and efficiency Some have even questioned if open-source LLM providers are not doing some shenanigans to provide the prices they offer. VC funded…
2024
- 22NT
With the latest launch from Google I've added support for Gemma 3 270M, the speed for local LLM to TTS token time is incredible! This is an heavy obvious work in progress - any contributions or tips would be welcome. The idea is to have a fast moving edge model playground, and maybe have some utility (like the e reader) on the side.
2025 · github.com
- 23IB
hey hn, I built an open-source Perplexity clone that can run local LLMs and cloud LLMs. It's fully self-hostable through Docker and uses ollama to support local LLMs. The demo video in the repository shows me running it locally with llama3 on my M1 Macbook Pro. I'm open to any suggestions or feedback, thanks!
2024 · github.com
- 24IB
Hey HN, I built a website where you can train Llama 3.1 8b & 70b (4bit) on your data. I use unsloth in the backend and the training is done on H100s which I rent programmatically from Runpod. I'd love some feedback. If you would be interested in using it feel free to book a chat with me: cal.com/hamada/tunellama-intro Happy to give you free credits :) P.S. I'm also looking for a co-founder as I have big plans for this.
2024 · tunellama.com
Ranked by how close each launch is in meaning, then by votes. Refine with a description →