I Put DeepSeek V4.1 Inside ChatGPT + Claude Code… It Works
DeepSeek V4.1 is live in experimental preview, and you can plug it into Codex CLI and Claude Code right now using my Relay-AI tool! In this video, we test the brand new multimodal vision capability, clock blistering speeds hitting up to 500 tokens per second, and look at the ultra-low Flash pricing. Note: this experimental preview is scheduled through September 10, so make sure to test it while it's active. Here is what we cover: 1. Setting up DeepSeek V4.1 in ChatGPT app and Claude Code 2. Running code reasoning and codebase queries 3. Testing the new vision analysis: feeding an image to see if it accurately identifies the object 4. Token speed and actual API usage costs Relay-AI Project on GitHub: https://github.com/jacob-bd/relay-ai ⏱️ Timestamps: 0:00 DeepSeek V4.1 Setup & Subscription Access 1:45 Loading the Model in Codex CLI 3:30 Codebase Description & Reasoning Speed 5:10 Testing Live Vision Analysis with macOS Webcam 7:05 500 Tokens/Sec & Ultra-Cheap Flash Pricing 📲 Daily AI news on Telegram → https://t.me/genaispot 📺 Subscribe to our YouTube Channel: https://www.youtube.com/@GenAISpotlight ✅ Follow GenAI Spotlight on TikTok: https://www.tiktok.com/@genai.spotlight #️⃣ Follow GenAI Spotlight on X: https://x.com/GenAISpotlight 👨🏽⚕️ Follow GenAI Spotlight on LinkedIn: https://www.linkedin.com/company/genai-spotlight #deepseek #codex #claudecode #aiagents #coding #vibecoding #developer