opencode - Tech blog

Tech blog

opencode

A collection of 19 posts

Local Harness Benchmark: Pi Coding Agent vs. OpenCode

In the rapidly evolving landscape of AI-assisted development, choosing the right "harness" is as critical as choosing the model itself. This benchmark explores two of the most prominent open-source harnesses for local LLMs: Pi Coding Agent and OpenCode. While both support a vast array of open-weight models, they

nca - 10x lighter alternative to opencode/claudecode written in rust

A lightweight, Rust-powered alternative to OpenCode/ClaudeCode that runs 10x lighter in terms of RAM usage. It's an early project but already functional with core features like tools and skills. What NCA Offers * Interactive TUI and line REPL — switch between full-screen interface and line-oriented shell * Detached sessions — run

OpenCode vs Pi: Which AI Coding Agent Should You Use?

A practical breakdown of the two most popular open-source AI coding agents in 2026. OpenCode Pi GitHub Stars 140K+ ~15K Made by AnomalyCo Mario Zechner Vibe Batteries included Build it yourself Language TypeScript TypeScript OpenCode is like VS Code out of the box. Pi is like Neovim. Both get the

A che punto siamo con le AI Locali? Gemma4-26B-A4B vs Qwen3.5-35B-A3B

intelligenza artificiale Featured

A che punto siamo con le AI Locali? Gemma4-26B-A4B vs Qwen3.5-35B-A3B

Si può fare Vibe Coding con LLM Locali? A seguire alcuni benchmark "real world" agentici con OpenCode (alias ClaudeCode opensource alternative) e Pi-Coding-Agent (alias il core harness di OpenClaw). Cosa testare? Ho creato una piccola applicazione Bun/Svelte e chiedo al LLM di fare alcune modifiche che richiedono

Are Local LLMs good enough for Vibe Coding? Gemma4-26B-A4B vs Qwen3.5-35B-A3B

opencode Featured

Are Local LLMs good enough for Vibe Coding? Gemma4-26B-A4B vs Qwen3.5-35B-A3B

Following some "real world" agentic benchmarks with OpenCode (aka ClaudeCode opensource alternative) and Pi-Coding-Agent (aka the OpenClaw core harness). What to test? I created a small Bun/Svelte application an I ask to the LLM to do some changes that require custom skill usage and custom CLI tool

The BEST Local LLM for opencode ! Gemma 4 26B A4B. No GPU required

Google's new Gemma 4 26B A4B is making waves in the local LLM community. This Mixture-of-Experts model only activates 4B parameters out of 26B, making it surprisingly fast on consumer hardware, even on a MacBook without a GPU. In a new video, the model is tested with opencode

Opencode vs Pi: Local LLM Benchmark Results

Opencode vs Pi: Local LLM Benchmark Results

Local LLMs are more powerful than many people realize. While smaller models are certainly less capable than larger online alternatives, they can leverage tools to identify errors and self-correct, significantly improving accuracy. For many use cases, local models offer a compelling option: no subscription fees and complete privacy since they

Opencode Benchmark Dashboard: La Guida Definitiva per Trovare i Migliori LLM Locali

intelligenza artificiale

Opencode Benchmark Dashboard: La Guida Definitiva per Trovare i Migliori LLM Locali

Negli ultimi anni, i modelli linguistici di grandi dimensioni (LLM) hanno fatto passi da gigante, e sempre più utenti cercano soluzioni che possano funzionare direttamente sul proprio computer, senza dipendere da servizi cloud. Opencode Benchmark Dashboard è uno strumento opensource che permette di confrontare le performance dei diversi modelli LLM

OpenCode Benchmark Dashboard: Find the Best Local LLM for Your Computer

opencode Featured

OpenCode Benchmark Dashboard: Find the Best Local LLM for Your Computer

If you're looking to run large language models (LLMs) locally, finding the right balance between accuracy and speed can be challenging. OpenCode Benchmark Dashboard is a new tool that helps you compare different LLM performances directly on your hardware. What is OpenCode Benchmark Dashboard? OpenCode Benchmark Dashboard is

George Hotz e il Futuro del Coding Agentico con openclaw e opencode

George Hotz, il leggendario hacker noto per aver creato il primo jailbreak dell'iPhone e fondatore di comma.ai e Tiny Corporation, è tornato con un nuovo video che esplora le frontiere più avanzate dell'intelligenza artificiale. In questo stream, registrato il 1° febbraio 2026, Hotz ci porta

GLM4.7-Flash the new Local LLM king at 30B A3B ?

GLM4.7-Flash represents Z.ai's latest breakthrough in the 30B parameter class, delivering a Mixture-of-Experts (MoE) model that balances high performance with efficiency. Released in January 2026, this model has quickly established itself as a formidable contender in agentic coding and general reasoning tasks. Technical Architecture Model Specifications:

opencode with superpowers. It can do everything in a container with docker and nix

opencode Featured

opencode with superpowers. It can do everything in a container with docker and nix

TL;DR: A lightweight Docker image (~223MB) that combines Nix package manager with non-root isolation, enabling AI coding assistants like OpenCode and Claude Code to install any software safely without admin privileges or home directory leaks. UPDATE: If you want a more light and mainstream approch you can use the

Vibe Coding Safely: The Ultimate Guide to AI Development with OpenCode and NixOS via docker-nixuser

opencode Featured

Vibe Coding Safely: The Ultimate Guide to AI Development with OpenCode and NixOS via docker-nixuser

In the era of AI-powered development, tools like OpenCode are revolutionizing how we write, test, and debug code. These intelligent assistants can generate code, execute commands, and even auto-correct errors, all from the comfort of your terminal. But with great power comes great responsibility. The Security Dilemma OpenCode is a

opencode tweaking: mastering SKILL.md vs AGENTS.md

opencode Featured

opencode tweaking: mastering SKILL.md vs AGENTS.md

As OpenCode continues to revolutionize how developers interact with AI coding agents, understanding the proper organization of project configuration files becomes crucial for maximizing productivity. In this guide, we'll explore the practical differences between SKILL.md and AGENTS.md files, and how to structure them effectively across global

Gli Agenti AI meritano una promozione nel 2026

opencode Featured

Gli Agenti AI meritano una promozione nel 2026

Alla fine del 2025, gli agenti AI hanno davvero preso vita. In parte perché i modelli sono migliorati, ma soprattutto perché gli abbiamo dato gli strumenti per portare la loro capacità oltre il puro ragionamento. Ora gli agenti di programmazione controllano il terminale, eseguono test per validare il loro lavoro,

Opencode Local LLM test with Nemotron-3-Nano-30B-A3B vs Qwen3-Coder-30B-A3B vs gpt-oss-20b-mxfp4

Tested the new Nvidia model, Nemotron 3 Nano 30B A3B, focusing on its performance in local coding and interaction tasks. The model was evaluated using a quantized version, and the tests were conducted in CPU mode due to hardware limitations. The user compared the model's performance with other

L'AI Non Garantisce il Successo: Verità sullo Sviluppo di Prodotti

In un'era dominata dall'entusiasmo per l'intelligenza artificiale, dove tutti promettono che "questa volta è diverso" e che l'AI cambierà ogni aspetto dello sviluppo software, c'è una voce fuori dal coro che merita attenzione. Dax, sviluppatore del progetto Open

Vibe Coding Myths vs. Reality: Why AI Isn’t a Bubble. Insights from llm-eval-simple

Vibe Coding Myths vs. Reality: Why AI Isn’t a Bubble. Insights from llm-eval-simple

Vibe Coding Myths and Reality: AI Isn't a Bubble AI could be a financial bubble but it isn't a technical bubble. With basic knowledge you can build and optimize almost without effort. With a bit of prompt engineering and an hour, I built Network Monitor -

OpenCode, usa il computer al posto tuo. Alternativa GRATIS a Claude Code

opencode Featured

OpenCode, usa il computer al posto tuo. Alternativa GRATIS a Claude Code

Openode è un innovativo strumento da riga di comando che rivoluziona l'interazione con il computer attraverso agenti intelligenti. Questo versatile strumento permette di automatizzare compiti complessi e di sfruttare l'intelligenza artificiale per eseguire comandi sul sistema, rappresentando un salto di qualità rispetto alle estensioni tradizionali di