Topic index

Browse every topic

121 articles organised by subject rather than format — so you can start from the model or the hardware you actually care about.

AI Video

ByteDance's cinematic model — 2.0 workflows, the 2.5 launch, prompts and every head-to-head.

Hailuo 3.0 — native 2K with built-in audio, the v2 API, omni-reference and open weights.

Veo 3.1 — photorealism, native dialogue, published per-second pricing and the Flow editor.

Kling 3.0, Turbo and Omni — motion control, native 4K and six cuts in one generation.

OpenAI Sora 2 — physics realism, cameos, native audio and the social-first workflow.

Gen-4 and the professional editing toolkit — inpainting, motion brush and director controls.

Building channels that never show a face — formats, Shorts, retention and monetization.

The craft layer that applies to every model — cost control, upscaling, consistency and image-to-video.

The complete-song generator — custom mode, lyric structure, stems and commercial rights.

Section-level control, remixing and inpainting for producers who want their song, not a song.

Voices, v3 audio tags, music tools and the only production-ready music API we ship through.

Lyria, generators, licensing risk and what is actually safe to monetize.

Voice cloning, text-to-speech and the offline stacks — Kokoro, XTTS, Piper, Chatterbox.

VRAM maths, RTX buying guides and what actually predicts local performance.

Llama, Qwen, DeepSeek, GLM, Kimi and Gemma — running frontier models on your own box.

The two runners most people start with — setup, model management and where each wins.

Wan, HunyuanVideo, LTX and ComfyUI — open-weight video on consumer hardware.

Whisper transcription and offline voice — Kokoro, Piper, XTTS and Chatterbox on your own machine.