1.1.0rc2

RubyLLM 1.1.0.rc2: Smart Error Handling & Testing Improvements 🛠️

This second release candidate introduces a new error handling paradigm for tools, improves error parsing across providers, and enhances testing infrastructure. The focus is on making applications more resilient and developer-friendly.

🛠️ Smarter Tool Error Handling

A completely new approach to handling errors in tools that lets LLMs handle recoverable issues while bubbling up critical errors to your application:

class Weather < RubyLLM::Tool
  description "Gets current weather for a location"
  param :latitude, desc: "Latitude (e.g., 52.5200)"
  param :longitude, desc: "Longitude (e.g., 13.4050)"

  def execute(latitude:, longitude:)
    validate_coordinates!(latitude, longitude)
    response = Faraday.get(weather_api_url(latitude, longitude))
    
    case response.status
    when 429
      # Return errors the LLM should know about and can retry
      { error: "Rate limit exceeded. Please try again in 60 seconds." }
    when 200
      JSON.parse(response.body)
    else
      # Let serious problems bubble up
      raise "Weather API error: #{response.status}"
    end
  end
end

When to Return vs Raise Errors

• Return errors to the LLM when:

  • Input validation fails
  • The operation can be retried (rate limits, temporary failures)
  • Alternative approaches might work

• Let errors bubble up when:

  • The tool encounters unexpected states
  • System resources are unavailable
  • Authentication or authorization fails
  • Data integrity is compromised

🔍 Provider Improvements

• Better error parsing for AWS Bedrock responses
• Enhanced system prompt handling across all providers
• with_instructions(replace: true) now available for Plain Old Ruby Objects

🧪 Testing Infrastructure

New testing capabilities focusing on error handling:

• Added chat_error_spec to catch problems with error parsing
• Enhanced CI workflow with prerelease version checking

📚 Documentation Updates

• New comprehensive model information guide listing 100+ models with capabilities
• Updated error handling guidelines
• Streamlined contribution guidelines

---

This is a release candidate. Please test thoroughly, especially the new error handling features!

gem 'ruby_llm', '1.1.0.rc2'

Contributors

Thanks to everyone who contributed to this release, especially @tpaulshippy for Bedrock improvements, and @keithrbennett for the model information guide.

New Contributors

• @keithrbennett made their first contribution in #71

Full Changelog: 1.1.0rc1...1.1.0rc2

1.1.0rc1

RubyLLM 1.1.0.rc1: Cloud Expansion & Better Conversations 🚀

We're excited to introduce RubyLLM 1.1.0.rc1, expanding cloud provider support and enhancing conversation management. This release candidate brings AWS Bedrock support, improved system prompts, and much more.

🌩️ AWS Bedrock Support

Access Claude models through your existing AWS infrastructure:

RubyLLM.configure do |config|
  config.bedrock_api_key = ENV.fetch('AWS_ACCESS_KEY_ID')
  config.bedrock_secret_key = ENV.fetch('AWS_SECRET_ACCESS_KEY')
  config.bedrock_region = ENV.fetch('AWS_REGION')
  config.bedrock_session_token = ENV.fetch('AWS_SESSION_TOKEN') # optional
end

# Using Claude through Bedrock
chat = RubyLLM.chat(model: 'claude-3-5-sonnet', provider: 'bedrock')

Great for teams who want to keep everything within their existing AWS environment.

🧠 Simpler System Instructions

We've introduced with_instructions for a cleaner way to set system prompts:

chat = RubyLLM.chat
  .with_instructions("You are a helpful coding assistant that specializes in Ruby")
  .ask("How would I implement a binary search tree?")

You can also explicitly replace previous instructions instead of accumulating them:

chat.with_instructions("New instructions", replace: true)

This works across all providers, with smart handling of differences between OpenAI, Anthropic, and others.

🔀 Smarter Model Resolution

Model identification is now much smarter:

• Model Aliases: Use simple names like gpt-4o instead of gpt-4o-2024-11-20
• Provider-Specific Matching: Target specific providers with the same base model
• Exact Match Priority: Exact matches are prioritized over aliases

# These all work now
chat = RubyLLM.chat(model: 'claude-3-5-sonnet')
chat = RubyLLM.chat(model: 'claude-3-5-sonnet', provider: 'anthropic')
chat = RubyLLM.chat(model: 'claude-3-5-sonnet', provider: 'bedrock')

🛠️ Tool Improvements

Tools get even better with support for parameterless tools:

class RandomNumber < RubyLLM::Tool
  description "Generates a random number between 1 and 100"
  
  def execute
    rand(1..100)
  end
end

🚂 Rails Enhancements

ActiveRecord integration improvements make Rails usage even better:

# Now all these methods return self for chainability
chat = Chat.create!(model_id: 'gpt-4o-mini')
  .with_instructions("You are a helpful assistant")
  .with_tool(Calculator)
  .ask("What's 123 * 456?")

🔧 Technical Improvements

• Fixed multimodal inputs for Bedrock Anthropic
• Improved system prompt handling across all providers
• Enhanced streaming functionality with better error parsing
• Fixed model refreshing for read-only filesystems in production
• Improvements to DeepSeek and Anthropic providers

📖 Documentation Updates

We've updated our documentation to cover:

• Working with multiple providers
• Using the model registry effectively
• Deploying in production environments

---

This is a release candidate. Please try it out and report any issues before the final release!

gem 'ruby_llm', '1.1.0.rc1'

Contributors

Thanks to everyone who contributed to this release, especially @tpaulshippy for the AWS Bedrock implementation, @kieranklaassen for ActiveRecord improvements, and @seuros and @redox for their contributions.

What's Changed

• Support tools without params by @redox in #62
• Switched from git ls-files to Ruby's Dir.glob for defining spec.file by @seuros in #84
• 16: Add support for Claude through Bedrock by @tpaulshippy in #50
• Fix refresh of Bedrock models and remove some expired credentials from cassettes by @tpaulshippy in #89
• fix acts_as delegation to return self instead of RubyLLM by @kieranklaassen in #82

New Contributors

• @redox made their first contribution in #62
• @seuros made their first contribution in #84
• @tpaulshippy made their first contribution in #50
• @kieranklaassen made their first contribution in #82

Full Changelog: 1.0.1...1.1.0rc1

1.0.1

🚀 Improvements & Bug Fixes

This release has some important fixes and quality-of-life improvements:

• Fixed temperature handling for OpenAI's o1 and o3 models (they require a temperature of 1.0, which we now normalize automatically)
• Better tool naming with proper stripping of unsupported characters for OpenAI
• Improved model capabilities detection for vision and function support
• Added base64 to dependencies for Ruby 3.4.0+ compatibility
• Enhanced Rails integration documentation with system message examples
• Added VCR support for testing (because integration tests shouldn't require real API calls)
• Fixed model refresh bug - no more stale model lists when you call RubyLLM.models.refresh!

📄 Documentation Enhancements

We've enhanced our documentation across the board:

• Added a section on system prompts to the chat guide
• Improved the contributing guidelines with model naming conventions
• Better examples of streaming responses in the README

🧪 Testing Improvements

• Added comprehensive specs for model filtering and chat functionalities
• Better CI/CD workflows for version validation

---

🏆 Shoutouts

Thanks to our contributors for this release: @dalthon, @jaggdl, @Thomascountz, @stevegeek, @jeduardo824, @redox, and @mhmaguire. This is what makes open source great.

New Contributors

• @mhmaguire made their first contribution in #59
• @Thomascountz made their first contribution in #60
• @jaggdl made their first contribution in #56
• @dalthon made their first contribution in #57
• @redox made their first contribution in #38
• @stevegeek made their first contribution in #28
• @jeduardo824 made their first contribution in #54

Full Changelog: 1.0.0...1.0.1

1.0.0

RubyLLM 1.0.0

A beautiful way to work with AI in Ruby. RubyLLM a unified interface to modern AI models from multiple providers with an elegant, Ruby-like API.

Features

• Unified API for OpenAI, Anthropic Claude, Google Gemini, and DeepSeek
• Simple conversation interface with automatic history tracking
• Consistent streaming that works the same way across all providers
• Built-in token tracking for cost management
• Tool integration with a clean, Ruby-like interface
• Rails integration with ActiveRecord persistence via acts_as_chat
• Multimodal support for images, PDFs, and audio
• Embeddings for vector search and semantic analysis
• Image generation with DALL-E and other providers
• Comprehensive error handling with specific error types
• Minimal dependencies for a lightweight footprint

Installation

# In your Gemfile
gem 'ruby_llm'

# Or install directly
gem install ruby_llm

Quick Start

# Configure with API keys
RubyLLM.configure do |config|
  config.openai_api_key = ENV['OPENAI_API_KEY']
  config.anthropic_api_key = ENV['ANTHROPIC_API_KEY']
  config.gemini_api_key = ENV['GEMINI_API_KEY']
end

# Start chatting
chat = RubyLLM.chat
chat.ask "What's the best way to learn Ruby?"

# Or generate an image
image = RubyLLM.paint("a sunset over mountains")
image.save("sunset.png")

# Or create embeddings
embedding = RubyLLM.embed("Ruby is a programmer's best friend")

Documentation

Full documentation is available at rubyllm.com

License

Released under the MIT License.

0.1.0.pre42

Full Changelog: https://github.com/crmne/ruby_llm/commits/0.1.0.pre42