Skip to content

Releases: mazen160/llmquery

v0.2

20 Sep 21:02
82c165c

Choose a tag to compare

LLMQuery Model Updates

Release Date: September 20, 2025

Overview

Comprehensive update of all LLM provider models.

What's New

Latest Models Added

Anthropic Claude

  • Claude 4.x Series (New Generation):
    • claude-opus-4-1-20250805 - Latest Opus with enhanced reasoning
    • claude-opus-4-20250514 - Claude 4 Opus baseline
    • claude-sonnet-4-20250514 - Claude 4 Sonnet for balanced performance
  • Claude 3.7:
    • claude-3-7-sonnet-20250219 - Extended thinking mode
    • claude-3-7-sonnet-latest - Latest 3.7 variant

OpenAI

  • GPT-4.1 Series:
    • gpt-4.1 - Enhanced instruction following, 1M context
    • gpt-4.1-mini - Compact version of GPT-4.1
  • Advanced Reasoning Models:
    • o3 - Next-generation reasoning model
    • o4-mini - Fast reasoning capabilities

Google Gemini

  • Gemini 2.5 Series (Current Top Tier):
    • gemini-2.5-pro - Most capable multimodal model
    • gemini-2.5-flash - Fast, cost-efficient
    • gemini-2.5-flash-lite - Lightweight variant
  • Gemini 2.0 Series:
    • gemini-2.0-flash - Balanced performance
    • gemini-2.0-flash-lite - Efficient processing

Mistral (Updated with Latest 2025 Models)

  • Frontier Models:
    • mistral-medium-2508 - Latest multimodal frontier model (Aug 2025)
    • codestral-2508 - Latest code generation model (Aug 2025)
  • Vision Capabilities:
    • pixtral-large-latest - Advanced vision understanding
    • pixtral-12b-latest - Efficient vision processing

Ollama (Comprehensive Model Support)

  • Meta Llama 4.x: llama4, llama4:16x17b, llama4:128x17b
  • Latest Open Models: llama3.3, qwen3, gemma3n, phi4
  • Specialized Models: llama3.2-vision, deepseek-r1

Enhanced Providers

GitHub AI Models - Dynamic Discovery

  • New Feature: Live model catalog integration
  • Auto-Updates: Automatically discovers new models via GitHub's catalog API
  • Fallback Support: Static model list when API unavailable
  • Proper Model IDs: Updated to official AzureML registry format

DeepSeek - Streamlined & Corrected

  • Fixed Endpoint: Corrected to official /v1/ API endpoint
  • Official Models Only: deepseek-chat, deepseek-reasoner

Cleanup & Deprecations

Removed Deprecated Models

  • Anthropic: claude-3-5-sonnet-20241022 (deprecated Oct 2025)
  • Google: gemini-1.5-pro, gemini-1.5-pro-latest (deprecated Sept 2025)
  • Mistral: 8 legacy models including ministral-3b-latest, open-mistral-nemo
  • DeepSeek: deepseek-coder (not officially supported)
  • GitHub Models: Legacy non-prefixed model names

Updated Defaults to Current Models

  • Mistral: mistral-small-latestmistral-medium-2508
  • Google Gemini: gemini-1.5-flashgemini-2.5-flash
  • GitHub Models: gpt-4o-miniopenai/gpt-4.1

By the Numbers

Provider Models Added Models Removed Net Change
Anthropic 5 new models 2 deprecated +3
OpenAI 4 new models 0 +4
Google Gemini 5 new models 2 deprecated +3
Mistral 2 latest models 8 legacy -6 (streamlined)
Ollama 25+ models 0 +25+ (new support)
DeepSeek 0 1 unofficial -1 (cleaned)
GitHub Models Dynamic catalog 17+ legacy Dynamic + cleaner

Total: 40+ new models added, 30+ deprecated models removed

Key Improvements

1. Dynamic Model Discovery

GitHub Models now automatically discovers available models, ensuring the provider stays current without manual updates.

2. Official Documentation Compliance

All models verified against official provider documentation to ensure reliability and support.

3. Cleaner Model Lists

Removed legacy and deprecated models to reduce confusion and improve user experience.

4. Better Default Models

Updated default models to current-generation options for optimal performance.

5. Corrected Endpoints

Fixed API endpoints to match official provider specifications.

Provider Coverage

Provider Status Models Special Features
Anthropic Claude Updated 12 models Claude 4.x series
OpenAI Updated 30+ models GPT-4.1, o3, o4-mini
Google Gemini Updated 8 models Gemini 2.5 series
Mistral Refreshed 8 models 2025 frontier models
Ollama Enhanced 25+ models Comprehensive local support
DeepSeek Cleaned 2 models Official models only
GitHub Models Enhanced Dynamic Auto-discovery

v0.1.18

12 Feb 13:29

Choose a tag to compare

Updates

  • Resolved a bug in the Anthropic provider integration.

v0.1.17

11 Feb 00:07

Choose a tag to compare

🚀 Release: [Version v0.1.17] - Enhancements & New Templates

🔧 New Features

✅ Configurable Template Paths via Environment Variables ([#9](#9))

  • Introduced support for defining template paths using environment variables.
  • Enables users to customize and override default paths without modifying code.
  • Enhances flexibility for diverse deployment environments and workflows.

✅ RawQuery Mode for Direct LLM Interactions ([#5](#5))

  • Added a RawQuery mode to allow direct interaction with the LLM, bypassing predefined templates.
  • Provides dynamic querying capabilities for users who require ad-hoc responses.

📜 New Templates Added

🏷 JIRA Ticket Generator ([#8](#8))

  • A structured template for auto-generating JIRA tickets.
  • Helps streamline ticket creation and maintain consistency.

✍ Grammarly-Assisted Writing ([#7](#7))

  • A template that integrates Grammarly for improved writing clarity, grammar, and tone.
  • Ideal for refining content before submission.

🛠 Improvements & Fixes

  • 📖 Updated documentation to reflect new features and templates.
  • ⚙️ Minor refactoring and optimizations for better maintainability.

🔍 Testing & Validation

  • ✅ Verified template path customization via environment variables.
  • ✅ Tested RawQuery for direct LLM interactions.
  • ✅ Validated JIRA and Grammarly templates with sample inputs.

🔗 Related Issues & Tickets:

  1. [#9 - Configurable Template Paths](#9)
  2. [#8 - JIRA Ticket Generator](#8)
  3. [#7 - Grammarly-Assisted Writing](#7)
  4. [#5 - RawQuery Mode](#5)

What's Changed

Full Changelog: v1.0.16...v0.1.17

v1.0.16

02 Feb 13:33

Choose a tag to compare

minor fix: #1

v1.0.15

02 Feb 10:35
4ccf8cf

Choose a tag to compare

Mistral + Github AI Integrations + various updates (#3)

* wip

* wip

* wip

Initial release

13 Jan 09:49

Choose a tag to compare

v0.1.11

v0.1.11