What Glevd does here
Glevd founded BenchLM and runs it: collecting benchmark results from their public sources, maintaining the scoring method, and writing the analysis on the blog. 89 articles carry this byline, the first from August 2025.
Rankings follow the published method, not the author's preference or a sponsor's. The editorial policy sets out how sourced facts, measured results and commercial relationships are kept apart, and how errors are corrected on the page.
Recent articles
- What OpenAI announced at DevDay, September 29, 2026
- Chat Completions still works. Your tools may not., September 25, 2026
- Prompt caching pays on the second call, September 25, 2026
- The evaluator must expose the failure, September 22, 2026
- Jev cannot write. That is the feature and the limit., September 22, 2026
- API credits expire, cap you and differ by provider, September 17, 2026
- Inference is the part of AI you pay for, September 17, 2026
- Valid JSON can still be wrong, September 14, 2026
- Mercury 2.5 upgrades the calls you don't see, September 11, 2026
- Deprecated means flagged, not broken, September 4, 2026
Contact
Corrections and questions go to hello@benchlm.ai. The public account is @glevd on X.