Moonshot AI

Kimi K3

• 9 independent or government tests
• 8 source documents

Independent and government tests

• ExploitBench arbitrary code execution (ACE) milestone · CAISI (NIST), government evaluator · Cyber

  kimi k3 achieved ace on 0/41 samples, whereas the most cyber-capable models achieved ace on 20/41 samples on average (figure 3).

  US CAISI (NIST) · 23 July 2026 · Read the source
• exploitbench · CAISI (NIST), government evaluator · Cyber

  kimi k3 achieves a score of 32%, whereas glm-5.2 achieves a score of 24% (figure 1).

  US CAISI (NIST) · 23 July 2026 · Read the source
• Bio Capabilities Index (BCI) · SecureBio, independent evaluator · Chemical, biological, radiological and nuclear

  kimi k3 is comparable to closed-weights models approximately eight months ago on securebio’s bio capabilities index (bci) score, which aggregates multiple biology benchmarks into a single comparable number, kimi scores a 142, placing it seventh among the models we have evaluated to date.

  SecureBio · 7 August 2026 · Read the source
• Human Pathogen Capabilities Test (HPCT) · SecureBio, independent evaluator · Chemical, biological, radiological and nuclear

  human pathogen capabilities test (hpct): 58.5 ± 9.7%.

  SecureBio · 7 August 2026 · Read the source
• Molecular Biology Capabilities Test (MBCT) · SecureBio, independent evaluator · Chemical, biological, radiological and nuclear

  molecular biology capabilities test (mbct): 58.4 ± 6.8%.

  SecureBio · 7 August 2026 · Read the source
• Virology Capabilities Test (VCT) · SecureBio, independent evaluator · Chemical, biological, radiological and nuclear

  virology capabilities test (vct): 46.7 ± 5.4%.

  SecureBio · 7 August 2026 · Read the source
• World Class Biology (WCB) · SecureBio, independent evaluator · Chemical, biological, radiological and nuclear

  world class biology (wcb): 55.0 ± 10.0%.

  SecureBio · 7 August 2026 · Read the source
• biotier-permit set · SecureBio, independent evaluator · Misuse & jailbreak

  kimi k3 correctly permits 99.3% of questions on our biotier-permit set.

  SecureBio · 7 August 2026 · Read the source
• biotier-refuse set · SecureBio, independent evaluator · Misuse & jailbreak

  kimi k3 refuses on only 26.9% of the questions in our biotier-refuse set.

  SecureBio · 7 August 2026 · Read the source

No document we read states: a release date, a safety framework, whether a capability level was crossed, a test by the provider, a safeguard. No incident on record names this version.

Some documents we hold are stored in lower case only, so their quotes are shown in lower case. The source link opens the original.

Compare with other versions