# AI Flash Report

> Frontier AI model tracker. Every major model release with benchmarks,
> pricing, context window, lineage, and primary-source coverage.
> Updated as models ship; news brief rebuilt daily at 05:00 UTC.

## Models
- [All tracked models](https://aiflashreport.com/models/): filterable hub of every frontier release
- [Released in the last 30 days](https://aiflashreport.com/model-releases.html#latest): deterministic recency block, models with a release date in the last 30 days
- [Monthly release pages](https://aiflashreport.com/model-releases/2026-06/): per-month launch lists at /model-releases/YYYY-MM/ (e.g. 2026-06, 2026-05, 2026-04), one page per month with 3+ releases
- [Model releases JSON API](https://aiflashreport.com/api/model-releases.json): machine-readable list of tracked model releases (identity fields), newest first
- [Claude Sonnet 4.6](https://aiflashreport.com/models/anthropic-claude-sonnet-4-6/): Anthropic, released 2026-02-17
- [Gemini 3.1 Pro](https://aiflashreport.com/models/google-gemini-3-1-pro/): Google, released 2026-02-19
- [GPT-5.3 Codex](https://aiflashreport.com/models/openai-gpt-5-3-codex/): OpenAI

## Benchmarks
- [ARC-AGI-2 leaderboard](https://aiflashreport.com/benchmarks/arc-agi-2/)
- [MMLU-Pro leaderboard](https://aiflashreport.com/benchmarks/mmlu-pro/)
- [SWE-bench Verified leaderboard](https://aiflashreport.com/benchmarks/swe-bench-verified/)
- [GPQA Diamond leaderboard](https://aiflashreport.com/benchmarks/gpqa-diamond/)
- [LiveCodeBench leaderboard](https://aiflashreport.com/benchmarks/livecodebench/)
- [HumanEval leaderboard](https://aiflashreport.com/benchmarks/humaneval/)

## News
- [Daily AI news brief](https://aiflashreport.com/news/): RSS + Reddit + Perplexity, deduplicated, 05:00 UTC

## Optional
- [Sitemap](https://aiflashreport.com/sitemap-index.xml)
- Data updates: model releases tracked in `data/model_releases.json` in the public repo
- Editorial: AI Flash Report editorial team. Citation OK with link to canonical URL.