Kimi K3 finds 32 of 86 vulnerabilities for just $19.37 in benchmark run
zeeg · x · 2026-07-23
Kimi K3 tops Warden-style benchmark on cost efficiency
A post claims Kimi K3 performed reliably on Warden’s benchmark, with no failures and no repair jobs during the run.
The attached chart compares models on a vulnerability-finding task:
- Kimi K3 found 32 / 86 known vulnerabilities (37.2%) at a total cost of $19.37
- Claude Sonnet 5 found 22 / 86 (25.6%) at $23.46
- Grok found 33 / 86 (38.4%) at $38.65
- Luna found 36 / 86 (41.9%) at $56.98
The poster says Kimi’s accuracy-to-price ratio is “amazing” and that they’re switching to it.
More from Models
- BullshitBench update: GPT-6-Astra beats all prior OpenAI models but still trails Anthropic — scaling01 · 2026-09-11
- Astra Scores 83% on GauntletBench, First Computer-Use Agent to Beat Human Baseline — ducha_aiki · 2026-09-11
- Kimi K2.8 Preview rolls out: near-K3 coding performance, 1M context for all tiers — teortaxesTex · 2026-09-11
- Looking for a classifier of software engineering task shapes to pick models per task — StewartalsopIII · 2026-09-11
- DeepSeek V4 Pro API to continue after Sept 2026, billing unchanged — teortaxesTex · 2026-09-11
- DeepSeek V4.1 Flash Hits 98% of GPT-6 Astra's Score at 1.4% of the Cost in Third-Party Benchmark — ayushtweetshere · 2026-09-11