Release notes for 0.23 schemas edition

[simonw]

42 commits! 0.22...bf80b8a

Almost all schema stuff, with the exception of:

• Upgrade to Pydantic v2 #775
• Prompt should be optional so model.prompt(attachments=llm.Attachment(...)) works #784
• Allow "reasoning" for extra-openai-models.yaml #766
• GPT-4.5 preview #795

[simonw]

Every referenced issue and PR:

• Fully migrate to Pydantic v2 #520
• Allow "reasoning" for extra-openai-models.yaml #766
• Add note about similarity function in "similar" command's doc #774
• Upgrade to Pydantic v2 #775
• Ability to specify a schema for structured output for models that support it #776
• llm prompt --schema X option and model.prompt(..., schema=) parameter #777
• Template support for schemas #778
• Support --schema filename in addition to --schema '{JSON}' #780
• llm schemas family of commands, similar to llm templates #781
• llm logs --schema X option #782
• Handle prompt: None better in logs #783
• Prompt should be optional so model.prompt(attachments=llm.Attachment(...)) works #784
• llm logs --schema, --data, --data-array and --data-key options #785
• Dedicated schemas page of documentation including a tutorial #788
• Implement a concise schema DSL alternative to JSON schema #790
• Implement --schema-multi option #791
• llm schemas dsl debug command #793
• Figure out which OpenAI models support schemas #794
• GPT-4.5 preview #795
• Add a note to llm models showing which models support schemas #796
• llm models --schemas option for listing models that support schemas #797
• Saving a schema as a template fails to preserve key order #798
• --schema t:people for loading the schema from a template #799
• llm logs --data-ids flag for returning conversation_id and response_id keys #800
• llm logs --id-gt and llm logs --id-gte options #801
• Show async as one of the features on llm logs --options #806

Calculated using:

git log 0eab3f5ff318c530105c9bb515cbd34a5d85c597..HEAD --pretty=format:"%B" | rg -o "#\d+" | sort -u | sed 's/^/- /'