v0.13.0

• Add new subcommands
  • photo - Generate photorealistic images
  • short - Provide a compact and focused answer
  • say - Convert text to speech using OpenAI's TTS model
  • config - Show currently applied settings
  • query - Query SQLite databases
  • edit - Edit prompt with your default editor
  • commit - Intelligently commit modified files based on current changes
• Add Google OCR feature powered by Gemini 3 Pro Preview
• Add Google Gemini image generation support
• Use prompt-based descriptive filenames for generated images
• Add configurable base URL support for all providers
• Add support for new models
  • GPT-5.1 models
  • GPT-5 Codex model
  • GPT Image 1.5 (new default for image generation)
  • Claude Opus 4.5
  • Gemini 3 preview models
  • Grok 2 Image
  • Perplexity and its Sonar models
  • Groq Qwen and DeepSeek
• Upgrade Claude default models to latest versions
• Show execution time in seconds instead of ms when > 10 seconds

v0.12.0

• Add new subcommands
  • image - Generate an image using GPT-5
  • transcribe - Transcribe text from an audio file using GPT-4o Transcribe
  • rewrite - Rewrite text with improved clarity and style
  • jq - Get help on using jq
  • svg - Generate an SVG image from a text prompt
• Add support for new GPT-5 models
• Add support for latest xAI models
• Update default models
• reply command: Improve prompt
• rename command: Support renaming several files and fix timestamp handling

v0.11.0

• Add new subcommands
  • fast - Shortcut for groq gemma2-9b-it
  • local - Shortcut for ollama llama3.2
  • value - Return only the value/answer without explanations
  • reply - Generate a reply to a conversation passed via stdin
  • docker - Use Docker development as the prompt context
  • git - Use Git development as the prompt context
  • nix - Use Nix development as the prompt context
• Add support for new OpenAI models
• Use v4 as the default models for Claude
• Add support for Google's Gemini models
  and use Gemma 2 as the default gemma model

v0.10.0

• Add support for DeepSeek
• Add support for Cerebras
• Add support for xAI's Grok
• Don't read stdin for commands which don't support it
• Don't move file to current directory when renaming in another directory
• Add subcommands for Typst (ty) and LilyPond (ly)
• Display used subcommand in status line
• Update Ollama default models
• Add --json-schema CLI flag
• rename: Always use a lowercase t for timestamp and improve prompt

v0.8.0

• Add new sub-command ocr to extract text from images
• Add new sub-command rename to rename files based on their content
• Add sub-commands with a programming language as the prompt context
• Add new sub-command changelog to generate a changelog from git commits
• Always use the latest versions of Anthropic's 3.5 models
• Update default Llama versions

v0.7.0

• Add flag --json to activate JSON output mode
• Add flag --raw to print the raw LLM response without any metadata
• Add more aliases for Anthropic, use new versions as default models
• CLI output: Show resolved model id instead of the used alias
• Add more aliases for OpenAI models

v0.6.0

• Add support for all Groq and Ollama models
• Make Groq Llama3 the new default model
• Add several new aliases (check cai help)

v0.5.0

• Support adding text to the prompt via stdin
• Support local Ollama server
• Add examples to help output

v0.4.0

• Add support for Anthropic's Claude models
• Support running several models at once with cai all
• Syntax highlight the output as markdown with bat

v0.3.0

• Add support for sub-commands to specify the API provider
• Set up proper CLI args handling powered by clap.rs