v0.5.0

What's Changed

Features

• Implement ai.testing module with fake model by @anbuzin in #243
• Convert ai.experimental_generate into an ai.stream counterpart by @anbuzin in #244
• Migrate to Gateway v4 by @anbuzin in #254
• Implement image generation by @anbuzin in #246
• Implement video generation by @anbuzin in #248
• Implement text-to-speech by @anbuzin in #249
• Implement embedding model support by @anbuzin in #255
• Implement speech to text by @anbuzin in #257
• Implement reranking by @anbuzin in #258
• Improve ToolRunner API by @msullivan in #253
• Support passing store=False to openai SDK for Responses by @msullivan in #264
• Instrument ai.ops with telemetry by @anbuzin in #265

Fixes

• Add a model_input_kind field to mirror result_kind by @msullivan in #251
• Add a cancelled check to hook's set_result call by @msullivan in #252
• Propagate StreamEnd.finish_reason to UI finish event by @jaygupta17 in #259
• Detect video media types in FilePart.from_bytes by @jaygupta17 in #260

New Contributors

• @jaygupta17 made their first contribution in #259

Full Changelog: v0.4.2...v0.5.0