Skip to content

v0.0.4

Latest

Choose a tag to compare

@kailashahirwar kailashahirwar released this 18 Aug 12:53

OpenTryOn v0.0.4 Release Notes

More video APIs, dual-path local models, and Qwen3.8 understanding

Release Date: 17 August 2026

OpenTryOn v0.0.4 expands the invoke layer shipped in v0.0.3 with new first-party video providers, dual-path (API + local) models, Qwen3.8 multimodal understanding, and clearer agent-facing integration docs. Same CLI / MCP / registry surface — more models show up automatically.

What's New

Video providers

Model Path CLI
LTX-2.5 Official API + local Diffusers ltx-2.5-api / ltx-2.5
MiniMax Hailuo 2.3 Official API only hailuo-2.3
Alibaba Wan DashScope API + local Wan 2.2 Diffusers wan-api / wan-2.2
Runway Gen-4.5 Official API only runway-gen4.5

Understanding — Qwen3.8

Model Path CLI
Qwen3.8-Max DashScope OpenAI-compatible qwen3.8-max
Qwen3.8-27B Local Transformers (Qwen/Qwen3.8-27B) qwen3.8

Native text + image + video → text. Thinking on by default; --reasoning-effort xhigh|medium|low on Max; --no-thinking on both paths. MCP: understand_qwen3_8_max, understand_qwen3_8.

Docs / DX

  • Agent guidelines: Path A (first-party API) vs Path B (local / HF / Ollama / LM Studio / Unsloth) — docs/docs/advanced/model-integration-guidelines.md
  • New API & local docs pages for LTX, Hailuo, Wan, Runway, Qwen3.8
  • env.template updated (LTX_*, MINIMAX_*, DASHSCOPE_* / QWEN_*, RUNWAYML_*, QWEN38_MODEL_ID, …)

Install / Upgrade

pip install -U opentryon
# optional local/GPU models (LTX-2.5, Wan 2.2, Qwen3.8, …):
pip install -U "opentryon[local]"

From source:

git checkout v0.0.4
pip install -e .

Quick examples

opentryon video-generate --model ltx-2.5-api --prompt "runway walk" --duration 8
opentryon video-generate --model hailuo-2.3 --prompt "runway walk" --duration 6
opentryon video-generate --model wan-api --prompt "runway walk" --duration 5
opentryon video-generate --model runway-gen4.5 --prompt "runway walk" --duration 5
opentryon understand --model qwen3.8-max --image garment.jpg --prompt "Describe this outfit."
opentryon understand --model qwen3.8 --image garment.jpg   # needs GPU + [local]

Docs

Notes

  • No breaking CLI/MCP changes vs v0.0.3 — new registry entries only.
  • Local LTX / Wan / Qwen need opentryon[local] (and often a recent Diffusers/Transformers); see each model’s docs page.
  • Full agent/tool surfaces for Qwen Max remain on DashScope; OpenTryOn exposes the understand path (+ chat() escape hatch).

Links