Kimi K3 reaches verified frontier status in BenchLM ranking

BenchLM’s July 2026 frontier index ranks the current top 10 models by provisional score, then separates verified positions from candidates based on exact-source evidence. Kimi K3 is listed as a verified frontier model with a score of 83, placing it #3 overall with 28 exact rows.

Claude Mythos 5 leads the overall table with a score of 87 but remains a frontier candidate, while Claude Opus 5 ranks #2 with a score of 85 and verified status. GPT-5.6 Sol, Claude Fable 5, GPT-5.6 Terra, and Claude Sonnet 5 are also inside the overall top 10 but remain candidates because category eligibility is still blocking verification.

BenchLM defines frontier membership as the ten highest provisional scores in a selected view, with verification requiring at least 8 exact-source benchmark rows and eligible coverage across two or more categories. The pricing view shows Claude Sonnet 5 as the lowest priced frontier API at $10 / 1M out, while the July 2026 baseline starts the site’s history for entries, exits, and former-frontier comparisons.