517: Plan First, Think Less: Save Tokens, Improve Code
From Merge Conflict by soundbite.fm
June 1, 2026 · 35 min
About this episode
The episode discusses AI avatars, text-to-speech deepfakes, and provides development advice on LLM thinking modes and token costs.
Episode 517 starts with a light chat about AI avatars and new text‑to‑speech deepfakes before diving into LLM “thinking” modes—what baked‑in planning actually does, why it multiplies token costs, and when it helps or hurts. James and Frank give concrete dev advice: try low‑thinking settings, use big models for creative planning then smaller ones to execute, leverage harnesses/system prompts, and beware quantized local models often do better without thinking. Follow Us Frank: Twitter , Blog , GitHub James: Twitter , Blog , GitHub Merge Conflict: Twitter , Facebook , Website , Chat on Discord Music : Amethyst Seer - Citrine by Adventureface ⭐⭐ Review Us ⭐⭐ Machine transcription available on http://mergeconflict.fm Support Merge Conflict
People in this episode
Hosts: James, Frank
Topics covered
- AI avatars
- text-to-speech deepfakes
- LLM thinking modes
- token costs
- development advice
- creative planning
- local models
Keywords
- AI
- deepfakes
- LLM
- token costs
- development advice
- creative planning
- local models
Mentioned in this episode
Organizations: Merge Conflict, Twitter, Blog, GitHub, Facebook, Discord
Books & works: Amethyst Seer - Citrine
More episodes of Merge Conflict
- 518: Windows is Back! New Microsoft AI Coding Models · June 8, 2026 · 47 min
- 516: Evolving Agent Session Management · May 25, 2026 · 43 min
- 515: Mini-LED, IPS, and the Great Monitor Rabbit Hole · May 18, 2026 · 44 min
- 514: Running Local LLMs in VS Code · May 11, 2026 · 56 min
- 513: Agents Over Chat: The Future of Developer Workflows · May 4, 2026 · 58 min
- 512: Does Matter Really Matter? · April 27, 2026 · 37 min
Explore listener stats, chart rankings, contacts and more on the Merge Conflict podcast page.