v0.2.254

📝 AiCore Release Notes – v0.2.254

Release Date: 2025-12-19
Previous Version: 0.2.253 → New Version: 0.2.254

---

🚀 New Features and Improvements

• OpenAI Model Enhancements:

  • Added new OpenAI model configurations with updated metadata support

• Observability Improvements:

  • Added polars schema definition for consistent type inference in LlmOperationCollector
  • Implemented environment toggle for JSON storage in collector for flexible data handling

🐛 Bug Fixes

• Observability Fixes:
  • Improved message content extraction functionality
  • Updated dashboard template for better data visualization

---

Full Changelog: [0.2.253...0.2.254](0.2.253...0.2.254)

Thank you to all contributors! Please upgrade to enjoy the latest features and improvements.

v0.2.253

📝 AiCore Release Notes – v0.2.253

Release Date: 2025-11-26
Previous Version: 0.2.252 → New Version: 0.2.253

---

🚀 New Features and Improvements

• MCP Image Support:
  • Added comprehensive image handling support across LLM providers (PR #33)
  • Enhanced image_to_base64 with URL support and added detect_image_type functionality
  • Added default_image_template for Anthropic provider
  • Updated tool call message content handling in Groq and Mistral providers to accept string or list formats

🐛 Bug Fixes

• Image Handling Fixes:
  • Fixed image template handling to use image_url with url field
  • Fixed tool call message output in base_provider and OpenAI provider
  • Added image content masking in LLM operation records for improved observability

---

Full Changelog: [0.2.252...0.2.253](0.2.252...0.2.253)

Thank you to all contributors! Please upgrade to enjoy the latest features and improvements.

v0.2.252

📝 AiCore Release Notes – v0.2.252

Release Date: 2025-11-25
Previous Version: 0.2.251 → New Version: 0.2.252

---

🚀 New Features and Improvements

• Gemini Model Enhancements:

  • Added extra_content to ToolCallSchema and implemented tool schema handling in GeminiLlm (Commit)

• Model Metadata Updates:

  • Added anthropic-claude-opus-4-5-20251101 to metadata (Commit)

• Observability Enhancements:

  • Added async session listing and refined date filtering in db queries (Commit)
  • Added async db queries and date filters to dashboard, optimized session loading, refactored fetch_df and add_day_col (Commit)

🐛 Bug Fixes

• LLM Fixes:
  • Preserved extra_content in tool calls and cleared signature cache (Commit)

---

Full Changelog: [0.2.251...0.2.252](0.2.251...0.2.252)

Thank you to all contributors! Please upgrade to enjoy the latest features and improvements.

v0.2.251

📝 AiCore Release Notes – v0.2.251

Release Date: 2025-11-18
Previous Version: 0.2.250 → New Version: 0.2.251

---

🚀 New Features and Improvements

• Model Enhancements:

  • Added gemini-3-pro-preview model to models metadata (Commit)

• Observability Enhancements:

  • Added async polars db query functionality (Commit)
  • Improved event loop handling (Commit)
  • Updated row conversion logic (Commit)

---

🐛 Bug Fixes

• No bug fixes in this release.

---

Full Changelog: [0.2.250...0.2.251](0.2.250...0.2.251)

Thank you to all contributors! Please upgrade to enjoy the latest features and improvements.

v0.2.250

📝 AiCore Release Notes – v0.2.250

Release Date: 2025-11-17
Previous Version: 0.2.240 → New Version: 0.2.250

---

🚀 New Features and Improvements

• Observability Enhancements:

  • Added lazy loading for session data in the dashboard (PR #32)
  • Added purge_corrupted option to polars_from_file method (Commit)
  • Added purge_corrupted parameter to constructor and fetch_df method (Commit)

• LLM and Model Enhancements:

  • Added Zai provider integration (Commit)
  • Added Zai-GLM-4.5-Air, Zai-GLM-4.5, and Zai-GLM-4.6 model metadata (Commit)
  • Added GLM-4.5-Flash model to CUSTOM_MODELS (Commit)
  • Added tool execution callback support to LLM and MCP client (Commit)
  • Added message template helpers and updated message content logic in base_provider (Commit)
  • Added default text and image templates and updated model list for OpenAI (Commit)
  • Added OpenAI-GPT-5.1 model with pricing and context window details (Commit)
  • Added GPT-5.1 to OPENAI_RESPONSE_API_MODELS list (Commit)

• Other Improvements:

  • Standardized prompt formatting and improved system message extraction (Commit)
  • Improved error handling for chunk data processing (Commit)
  • Improved error logging with full tracebacks (Commit)
  • Improved system message handling in completion_args (Commit)
  • Updated chunk file retrieval to include subdirectories (Commit)
  • Changed chunk file sorting key from int to str for proper ordering (Commit)
  • Renamed .env-example to .env.example (Commit)

---

🐛 Bug Fixes

• LLM and Logger Fixes:

  • Prevented logging empty messages in providers (Commit)
  • Improved error handling for chunk data processing (Commit)

• Observability Fixes:

  • Added fallback file save on chunk write failure (Commit)

• Server Manager Fixes:

  • Ensured proper serialization of arguments and output in tool calls (Commit)

• Collector Fixes:

  • Improved system message handling in completion_args (Commit)
  • Updated chunk file retrieval to include subdirectories (Commit)
  • Changed chunk file sorting key from int to str for proper ordering (Commit)

• Metadata Fixes:

  • Added OpenAI-GPT-5.1 model with pricing and context window details (Commit)

---

Full Changelog: [0.2.240...0.2.250](0.2.240...0.2.250)

Thank you to all contributors! Please upgrade to enjoy the latest features and improvements.

v0.2.240

📝 AiCore Release Notes – v0.2.240

Release Date: 2025-10-17
Previous Version: 0.2.230 → New Version: 0.2.240

---

🚀 New Features and Improvements

• Observability Dashboard Enhancements:

  • Added time range filter, improved selection state, and enhanced UI (PR #29)
  • Refactored observability to use orjson for polars file loading (PR #29)
  • Implemented chunked session-based storage and updated polars export (PR #31)
  • Improved DB pooling and added async bulk insert (PR #31)
  • Added utils for LLM stats aggregation and session data deletion (PR #31)
  • Enhanced SQL Server support with MARS connection settings and improved async session handling (PR #31)
  • Updated model metadata with new haiku model and adjusted pricing details (PR #31)

• Claude Plans Support:

  • Added support for optional API key and access token, enhanced system prompt and config handling (PR #30)
  • Added OAuth authentication module and example script for access token retrieval (PR #30)

• LLM and Model Enhancements:

  • Supported full and partial dynamic pricing strategies (Commit)
  • Added dynamic pricing to sonnet-4 models (Commit)
  • Added strategy field to DynamicPricing with default (Commit)
  • Supported avoid_dynamic pricing and context window (Commit)
  • Added 1m ctx beta header and updated sonnet-4 metadata (Commit)
  • Added Anthropics beta context setter and updated pricing logic (Commit)

---

🐛 Bug Fixes

• Observability Fixes:

  • Handled list system prompts in LlmOperationRecord (Commit)
  • Implemented chunked session-based storage and updated polars export (PR #31)

• LLM Fixes:

  • Corrected context_window assignment and cleaned up spacing (Commit)

• MCP Fixes:

  • Supported both type and transport_type in server config (Commit)

---

Full Changelog: [0.2.230...0.2.240](0.2.230...0.2.240)

Thank you to all contributors! Please upgrade to enjoy the latest features and improvements.

v0.2.230

📝 AiCore Release Notes – v0.2.230

Release Date: 2025-09-30
Previous Version: 0.2.221 → New Version: 0.2.230

---

🚀 New Features and Improvements

• OpenAI Responses Support:

  • Added support for OpenAI responses API and tool streaming logic (PR #25)
  • Updated tool call schema and added message validation (PR #25)
  • Expanded tool call arguments and updated execution logic (PR #27)
  • Added tool call argument conversion methods and updated providers (PR #27)

• LLM Enhancements:

  • Refactored class/static methods to instance methods (PR #26)
  • Improved handling of ResponseCompletedEvent and stream output (PR #25)
  • Cleaned up normalize_responses and added verbosity argument (PR #25)
  • Updated reasoning model argument handling for max tokens (PR #25)

• Model Updates:

  • Added metadata for claude-sonnet-4-5-20250929 (Commit)

---

🐛 Bug Fixes

• OpenAI Responses:

  • Fixed tool call schema for responses API compatibility (PR #25)
  • Handled ResponseFunctionToolCall in tool call stream (PR #27)

• LLM Fixes:

  • Fixed _handle_tools to use instance method and self (PR #25)
  • Added messages fallback and commented out unused prompts (PR #25)

---

Full Changelog: [0.2.221...0.2.230](0.2.221...0.2.230)

Thank you to all contributors! Please upgrade to enjoy the latest features and improvements.

v0.2.221

📝 AiCore Release Notes – v0.2.221

Release Date: 2025-09-27
Previous Version: 0.2.22 → New Version: 0.2.221

---

🔧 Dependency & Version Updates

• Relaxed aiofiles requirements to ensure broader compatibility and flexibility in the project's dependency management.

---

Full Changelog: [0.2.22...0.2.221](0.2.22...0.2.221)

v0.2.22

📝 AiCore Release Notes – v0.2.22

Release Date: 2025-09-25
Previous Version: 0.2.21 → New Version: 0.2.22

---

🚀 New Features & Improvements

• Observability Enhancements

  • Added asynchronous record handling and file storage for improved logging and monitoring.
  • Refactored logger to use sys.stdout for output and disable file logging for more efficient, non-blocking operations.

---

🔧 Dependency & Version Updates

• Added aiofiles and orjson to requirements-dashboard to support new observability features.

---

Full Changelog: [0.2.21...0.2.22](0.2.21...0.2.22)

v0.2.21

📝 AiCore Release Notes – v0.2.21

Release Date: 2025-09-22
Previous Version: 0.2.2 → New Version: 0.2.21

---

🐛 Bug Fix

• Claude Opus Configuration

  • Fixed an issue with the max_tokens parameter to ensure correct handling and alignment with provider limits.

---

🔧 Dependency & Version Updates

• Minor internal adjustments.
• Bumped project version to 0.2.21.

---

Full Changelog: [0.2.2...0.2.21](0.2.2...0.2.21)