v1.3.0

Release v1.3.0

Changes since v1.2.0

• chore: bump to 1.3.0
• feat: skip transform for native multimodal models (gpt, claude, gemini, qwen3.5/3.6, kimi-k2.5/2.6, minimax-m3, glm-v etc)

v1.2.0

Release v1.2.0

Changes since v1.0.2

• ci: auto-delete existing release before creating new one
• ci: use commit messages as release notes body instead of github defaults
• ci: add auto-release workflow on tag push
• feat: image sequence numbering with content-hash dedup
• feat: add MiniMax VLM backend support
• 1.0.2