# Changelog

All notable changes to the LLM Model Comparison dataset will be documented in this file.

Format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).

## [Q1 2026 v2] - 2026-02-18

### Changed

- Major update: replaced all models with latest generations
- Added GPT-4.1 and GPT-4.1 mini (1M context window)
- Added o3 and o4-mini reasoning models
- Added Claude 4.5 family (Sonnet, Haiku, Opus)
- Added Gemini 2.5 Pro and Gemini 2.5 Flash
- Added Llama 4 Scout and Llama 4 Maverick (10M context window)
- Added DeepSeek V3 (MIT, open source)
- Added Mistral Large 3 (Apache 2.0, open source)
- Added Command A (Cohere)

### Removed

- Removed deprecated models: GPT-4o, Claude 3.5 family, Gemini 2.0/1.5, Llama 3.x, older Mistral and Cohere models

## [Q1 2026] - 2026-02-15

### Added

- Initial release with 16 LLM models across 7 providers
- 22 fields per model including benchmark scores, latency metrics, and API feature flags
- Providers: OpenAI, Anthropic, Google, Meta, DeepSeek, Mistral AI, Cohere

---

Published by [Salt Technologies AI](https://www.salttechno.ai)