Skip to content

v1.0.29

Latest

Choose a tag to compare

@github-actions github-actions released this 10 Nov 14:01
ae29aae

Highlights

  • Add MiniMax-H3-Image support to mmx image generate, including text-to-image (--aspect-ratio/--area or --size) and image editing with --image (1–16 references).
  • Support 768P resolution for MiniMax-H3 video generation.
  • mmx agent setup now keeps only MiniMax-M3.1 and MiniMax-M3, adds Zed setup, and defaults pay-as-you-go keys to MiniMax-M3.1-Flash. The model id is renamed from MiniMax-M3.1-Flash-Preview to MiniMax-M3.1-Flash.
  • Fix OAuth refresh coalescing and resource URL scoping, stop polling sleeps at the remaining deadline, and choose speech SDK sample rates from the audio format.

What's Changed

  • fix: coalesce concurrent automatic OAuth refreshes by @ch-z-hc in #291
  • fix: stop polling sleeps at the remaining deadline by @ch-z-hc in #283
  • fix: scope OAuth resource URL to OAuth credentials by @ch-z-hc in #284
  • feat(agent): keep only M3.1 and M3 in agent setup models by @SaladDay in #293
  • fix(agent): rename model id to MiniMax-M3.1-Flash by @SaladDay in #294
  • fix(agent): send effort for omp M3.1 Flash low/medium/high by @SaladDay in #295
  • feat(agent): add Zed agent setup by @SaladDay in #297
  • fix(agent): declare MiniMax native effort levels and no temperature for Pi and OpenCode by @SaladDay in #298
  • feat(agent): default pay-as-you-go keys to MiniMax-M3.1-Flash by @SaladDay in #296
  • fix: choose speech SDK sample rate from audio format by @Ricardo-M-L in #305
  • feat(video): support 768P resolution for MiniMax-H3 by @gamebody in #303
  • feat: support MiniMax-H3-Image generation and editing by @jellyjelly814 in #306

New Contributors

Full Changelog: v1.0.28...v1.0.29