Apr 4
Loading video player...
I built AnyModel — an open-source tool that lets you run Claude Code with ANY AI model. GPT-5.4, DeepSeek R1, Gemini 3.1, Codex, Llama, Qwen — even free models at $0 cost. Two terminals. Any model. One npm package. 🔗 Links: → Website: https://anymodel.dev → npm: https://npmjs.com/package/anymodel → GitHub: https://github.com/anton-abyzov/anymodel → GitHub: https://github.com/antonoly/claude-code-anymodel → NPM: https://www.npmjs.com/package/anymodel → Free OpenRouter key: https://openrouter.ai/keys 🎙️ Wispr Flow (voice-to-text I use for coding): https://wisprflow.ai/r?ANTON691 ⏱️ Timestamps: 0:00 — The problem: Claude Code locked to one provider 0:14 — Two terminals. Any model. I built this. 0:55 — How AnyModel proxy works (architecture) 1:32 — Architecture diagram 1:40 — Demo: DeepSeek R1 preset 2:26 — Why OpenRouter? 600+ models 2:45 — Google Gemma 4 (released today) 3:28 — How to get your free OpenRouter API key 3:57 — DeepSeek R1 thinking demo (127 × 389) 4:17 — Like & Subscribe reminder 4:37 — VS Code integration 4:55 — Codex preset + building a web calculator 5:21 — SpecWeave: structured AI development 5:47 — Verified Skills registry (100K+ skills) 7:31 — Calculator result: frontend + backend working 8:00 — Running Claude directly (npx anymodel claude) 9:06 — Free models: Nemotron from NVIDIA 9:33 — Multiple proxies on different ports 10:17 — Qwen3 Coder (free) + Gemma 4 11:00 — Agent Swarm with SpecWeave team-lead 11:54 — Security, performance & brainstorm agents 13:23 — Consolidated decision matrix from agents 14:29 — Fully offline with Ollama 14:49 — Llama 3.1 8B local demo 16:00 — Get started + closing 📦 Quick start: Terminal 1: OPENROUTER_API_KEY=sk-or-v1-... npx anymodel proxy deepseek Terminal 2: npx anymodel 🏷️ Built-in presets: • gpt → openai/gpt-5.4 • codex → openai/gpt-5.3-codex (coding) • gemini → google/gemini-3.1-flash-lite • deepseek → deepseek/deepseek-r1-0528 • mistral → mistralai/devstral-2512 (coding) • gemma → google/gemma-4-31b-it • qwen → qwen/qwen3-coder:free (FREE) • nemotron → nvidia/nemotron-3-super-120b (FREE) • llama → meta-llama/llama-3.3-70b (FREE) 🔧 Works with: → OpenRouter (200+ cloud models) → SpecWeave: https://spec-weave.com → Verified Skills: https://verified-skill.com → Ollama (fully local/offline) → Any OpenAI-compatible API 🔗 Links: → Website: https://anymodel.dev → npm: https://npmjs.com/package/anymodel → GitHub: https://github.com/anton-abyzov/anymodel → Free OpenRouter key: https://openrouter.ai/keys → SpecWeave: https://verified-skill.com/docs → Verified Skills: https://verified-skill.com #claudecode #anymodel #aitools #gptvsclaude #deepseek #openrouter #ollama #freeLLM #codingwithAI #aicoding 🔔 Subscribe for AI, automation, and startup content: https://youtube.com/@AntonAbyzov 🔗 Connect: 🐦 Twitter/X: https://x.com/aabyzov 🧵 Threads: https://threads.net/@aabyzov 💼 LinkedIn: https://linkedin.com/in/antonabyzov 🌐 Website: https://antonabyzov.com 👥 Community: 💬 Discord: https://discord.gg/UYg4BGJ65V 📱 Telegram: https://t.me/antonaipower 🎙️ Wispr Flow (voice-to-code I use daily): https://wisprflow.ai/r?ANTON691 🎯 Topics: AI engineering, automation, business growth, software practices.