Klingit

Developer Spent $8,857 on Claude Code — Lessons Learned

A developer documented a 14-day experiment using Claude Code (Opus 4.8, 1M context) across six projects, spending $8,857.62 for 3.884 billion tokens and 47,235 API requests. The post breaks down project-level costs (LightCraft V2 ~$4,200; AI news video pipeline ~$1,800; NZ WHV slot grabber ~$1,100, etc.) and highlights that prompt/context caching dominated token usage (1.322B cache writes; 2.499B cache hits; 86.4% hit rate), dramatically reducing marginal cost because cache hits are billed at ~1/10th of new input. The author contrasts Opus (better for architectural/judgment tasks) with Sonnet (cheaper for grunt work), describes configuration levers (settings.json effortLevel = "xhigh", CLAUDE.md behavioral constraints, PreToolUse/PostToolUse hooks), and offers practical lessons about AI blind spots (legacy stacks, platform policy limits, user-facing edge cases).

Agent-built generative video pipeline using Claude Code

A developer describes building a two-minute video entirely via an agentic Claude Code session (named “Simona”) that created and composed image generation, text-to-speech, AI-video, and ffmpeg editing skills. The post is a technical walkthrough showing how the agent iteratively built reusable "skills" (with SKILL.md docs and CLI wrappers), tracked costs in a WORKLOG.md ledger, and recovered after a git mishap that deleted assets. The author lists the models and services used (OpenAI gpt-image-2, Google Gemini/Nano Banana, Seedance 2.0, Kling, LTX, ElevenLabs, Google TTS, local Kokoro), provides a cost breakdown ($27.76 for the final locked cut; $45.26 total project spend), and documents engineering patterns and guardrails for safe agent-driven media production.

Avataar launches Varya, low-cost India-focused video AI

Avataar AI, a Peak XV-backed startup chosen for India’s India AI Mission, has released Varya, a video-generation model optimized for Indian cultural contexts and e-commerce use cases. Built by distilling Alibaba’s Wan 2.2 into a leaner model, Varya produces video much faster (a 5-second 720p clip in 45 seconds on an NVIDIA H200 versus 1,230 seconds for Wan 2.2) and at substantially lower cost. Avataar plans to price hosted output at ₹0.48 ($0.005) per second and will publish Varya’s weights and training data on the government’s India AI Kosh portal for self-hosting and modification. The company says the model was trained on curated data to better represent Indian festivals, clothing, food and architecture, and it is open to partnerships with video tools and enterprise customers.