# LLM Statistics

> Citable, dated statistics on AI models — pricing, releases, context windows, benchmarks, open source, local hardware, and market share — generated from BenchLM's live dataset. Updated September 10, 2026.

- [LLM Pricing Statistics (2026)](https://benchlm.ai/stats/llm-pricing) ([md](https://benchlm.ai/md/stats/llm-pricing.md)): Frontier LLM token prices are 84% below their March 2023 level as of September 2026, per the BenchLM Token Price Index (16 on a base of 100).
- [AI Model Release Statistics (2026)](https://benchlm.ai/stats/model-releases) ([md](https://benchlm.ai/md/stats/model-releases.md)): BenchLM tracked 198 AI model releases in the 12 months ending 2026-09-10 — a notable model release roughly every 2 days.
- [LLM Context Window Statistics (2026)](https://benchlm.ai/stats/context-windows) ([md](https://benchlm.ai/md/stats/context-windows.md)): As of September 10, 2026, the largest LLM context window tracked by BenchLM is 10M tokens, offered by Pokee AI's Pokee-Isaac 28B.
- [LLM Benchmark Statistics (2026)](https://benchlm.ai/stats/benchmarks) ([md](https://benchlm.ai/md/stats/benchmarks.md)): As of September 10, 2026, BenchLM tracks 435 LLM benchmarks across 10 categories for 489 AI models.
- [Open Source LLM Statistics (2026)](https://benchlm.ai/stats/open-source-llm) ([md](https://benchlm.ai/md/stats/open-source-llm.md)): As of September 10, 2026, 10 of the top 50 AI models on BenchLM (20%) are open-weight.
- [Local LLM Hardware Statistics (2026)](https://benchlm.ai/stats/local-llm-hardware) ([md](https://benchlm.ai/md/stats/local-llm-hardware.md)): As of September 10, 2026, BenchLM documents license, quantization, and reference hardware for 12 ranked open-weight models.
- [LLM Market Share Statistics (2026)](https://benchlm.ai/stats/llm-market) ([md](https://benchlm.ai/md/stats/llm-market.md)): As of September 10, 2026, the #1 AI model overall on BenchLM is Anthropic's Claude Fable 5.1, with an overall score of 84.61/100.

Canonical page: https://benchlm.ai/stats