Free, open-source AI video generator for Mac. Run LTX-2.5 and MiniMax H3 (Hailuo) locally on Apple Silicon: text-to-video and image-to-video with sound, lip-sync, character LoRA training, songs from lyrics (YuE2) and image editing (Qwen-Image-Edit). One-click install via Pinokio. Offline: no cloud, no API key, no NVIDIA GPU.
Python ai-filmmakingai-videoai-video-generatorapple-siliconhailuoimage-to-videolip-synclocal-ailora-trainingltx-2ltx-2-5ltx-videomacosminimax-h3mlxmusic-generationoffline-aipinokiotext-to-videovideo-generation
fix(agent): bump engine timeout 5min → 15min
mrbizarrobd37a3ed0ac1
diff --git a/agent/engine.py b/agent/engine.pyindex cb5157c..11479cd 100644--- a/agent/engine.py+++ b/agent/engine.py@@ -17,10 +17,13 @@ import urllib.requestfrom dataclasses import asdict, dataclass, field-# 5 minutes. Local models can take 30+ s to first token on a cold cache;-# the first request after spawning mlx-lm.server is the slowest. Keep-# this generous so an under-warmed model doesn't fail the first turn.-DEFAULT_TIMEOUT_S = 300+# 15 minutes. Local reasoning-class models (Qwen 3.6, DeepSeek R1) on+# long contexts can easily take 5-10 minutes per turn — they emit+# 1500-3000 tokens of chain-of-thought BEFORE the answer. Combined with+# a multi-shot session that has 50+ messages of history, inference is+# slow. The earlier 5-min cap was hitting in the middle of legitimate+# turns; user saw "engine error: timed out" mid-batch.+DEFAULT_TIMEOUT_S = 900@dataclass