Hackernews posts about Llama 4

1. Benchmarking rolvsparse on DeepSeek-R1 and Llama 4 – up to 82x vs. cuBLAS (rolv.ai)
   2 points by heggenhougen 29 days ago | 1 comments
2. Show HN: ROLV – 20x faster MoE FFN inference on Llama 4 Maverick vs. cuBLAS (rolv.ai)
   1 points by heggenhougen 30 days ago | 1 comments
3. CostRouter – Cut AI API costs 60% by routing to the cheapest capable model
   4 points by alex_1002 27 days ago | discuss
4. Show HN: 1k desktop beats vendor sparse library 474× on Mistral-7B (rolv.ai)
   1 points by heggenhougen 22 days ago | 1 comments
5. Show HN: UberSKILLS – Open-source Workbench for building AI agent SKILLS (github.com)
   1 points by felltrifortence 26 days ago | discuss
6. Tessera – 32 OWASP security tests for GPT-4o, Claude, Gemini, Llama 3 (github.com)
   1 points by alysheaib 15 days ago | discuss
7. Launch HN: RunAnywhere (YC W26) – Faster AI Inference on Apple Silicon (github.com)
   240 points by sanchitmonga22 29 days ago | 153 comments
8. Show HN: Experiments Mapping the "Primitive Layer" in Language Models (github.com)
   2 points by dchisholm125 24 days ago | discuss
9. Fixed a llama.cpp bug silently disabling Vulkan GPU on all 32-bit ARM devices
   3 points by perinban 5 days ago | discuss
10. Llama.cpp with CUDA Support on Original Jetson Nano (4GB) (github.com)
    2 points by Abishek_Muthian 11 days ago | 2 comments
11. Show HN: A single CLI to manage llama.cpp/vLLM/Ollama models (github.com)
    2 points by everlier 26 days ago | 1 comments
12. Show HN: Running LLM on smartwatch – found llama.cpp loading model twice in RAM
    1 points by perinban 6 days ago | discuss
13. Show HN: LiteParse, a fast open-source document parser for AI agents (github.com)
    12 points by freezed8 19 days ago | discuss
14. Show HN: Ava – AI Voice Agent for Traditional Phone Systems(Python+Asterisk/ARI) (github.com)
    7 points by hkjarral 27 days ago | 1 comments
15. Show HN: ClawMem – Open-source agent memory with SOTA local GPU retrieval (github.com)
    5 points by yoloshii 18 days ago | discuss
16. Tq-KV – Rust implementation of TurboQuant that works on GGUF models
    3 points by onurgokyildiz 7 days ago | discuss
17. Show HN: NVFP4 on Desktop Blackwell – 122B MoE on a Single RTX PRO 6000 31 tok/s
    2 points by jcartu 30 days ago | discuss
18. Show HN: WayInfer – Native GGUF engine that runs models larger than your RAM
    1 points by ahmedm24 6 days ago | discuss
19. Show HN: FC-Eval – CLI to Benchmark Local or Cloud LLMs on Function Calling (github.com)
    3 points by gauravvij137 22 days ago | discuss
20. Diffusion LLM may make most of the AI engineering stack obsolete
    3 points by victorpiles99 28 days ago | discuss
21. Show HN: The Common Infrastructure for Agentic Communication (cyrisai.dev)
    1 points by krishnamzg 26 days ago | discuss
22. April 2026 TLDR Setup for Ollama and Gemma 4 26B on a Mac mini (gist.github.com)
    325 points by greenstevester 5 days ago | 121 comments
23. Show HN: MicroSafe-RL – Deterministic 1.18µs safety layer for Edge AI (github.com)
    1 points by DREDREG 4 days ago | discuss
24. Show HN: Beta-Claw – I built an AI agent runtime that cuts token costs by 44% (github.com)
    2 points by Kyoiske 30 days ago | discuss
25. Show HN: OpenCastor Agent Harness Evaluator Leaderboard (craigm26.github.io)
    3 points by craigm26 16 days ago | 1 comments
26. Show HN: StackStats – Analytics tool for Substack writers, runs 100% locally
    1 points by rishikeshs 22 days ago | discuss
27. Show HN: An offline first focused agentic CLI application powered by Ollama
    4 points by g023 14 days ago | discuss
28. Show HN: Ollama-client-rs, a Rust client for Ollama (github.com)
    3 points by fushji about 19 hours ago | discuss
29. Show HN: Chat.nvim v1.4.0 – OpenClaw-like AI assistant for Neovim (github.com)
    3 points by wsdjeg 26 days ago | discuss
30. Show HN: SycoFact 4B: Open model detecting sycophancy and delusion confirmation (huggingface.co)
    2 points by iwalton3 9 days ago | discuss