Model profile
Kimi K2.5
Moonshot AIdeveloper
2026-01-01release date
#123 / 305overall rank
27eval lineages
Evidence summary
Kimi K2.5 has an estimated overall rank of #123; its 90% source-sensitivity interval is #72–#207. Its behavior-only rank is #122; company governance moves the combined estimate to #123. Published evidence spans 27 evals and 7 of 7 behavior components. Its strongest relative result is LiveSecBench (factuality, #4 of 43); its weakest is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #232 of 241).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗kimi-k2-5
- Hugging Face ↗moonshotai/Kimi-K2.5
- OpenRouter ↗moonshotai/kimi-k2.5
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
Published eval results
Rank is within that sub-eval. Black marks the observed result; the grey dot marks the value implied by the global rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #57 / 327 | ↓0.4996 | Source ↗official | |
| AgentAbstainabstain | #13 / 17 | ↑52 | Source ↗official | |
| AgentAbstaincar | #11 / 17 | ↑53.2 | Source ↗official | |
| AgentAbstainpaired | #16 / 17 | ↑33.4 | Source ↗official | |
| Arena Factuality — Text Arena (factuality-only weighting)factuality_bt_rating | #77 / 112 | ↑1423.0 | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #35 / 106 | ↑0.415 | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #37 / 44 | ↓90.9 | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #43 / 49 | ↓93.2 | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #44 / 50 | ↓75.9 | Source ↗official | |
| CAIS Risk Indexmachiavelli | #36 / 46 | ↓93.5 | Source ↗official | |
| CAIS Risk Indexmask | #25 / 52 | ↓18.5 | Source ↗official | |
| CAIS Risk Indextextquests_harm | #34 / 49 | ↓20.8 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #41 / 104 | ↓24.13 | Source ↗official | |
| DystopiaBenchbasaglia_score | #20 / 50 | ↓59.1 | Source ↗official | |
| DystopiaBenchbaudrillard_score | #17 / 50 | ↓44.77 | Source ↗official | |
| DystopiaBenchhuxley_score | #20 / 50 | ↓62.6 | Source ↗official | |
| DystopiaBenchlaguardia_score | #20 / 50 | ↓62.37 | Source ↗official | |
| DystopiaBenchorwell_score | #19 / 50 | ↓56.97 | Source ↗official | |
| DystopiaBenchpetrov_score | #24 / 50 | ↓72.53 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #112 / 241 | ↑16.28 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #232 / 241 | ↑66.83 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #32 / 241 | ↑94.44 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #117 / 239 | ↑96.36 | Source ↗official | |
| FORTRESSaverage_risk_score | #29 / 49 | ↓41.09 | Source ↗official | |
| FORTRESSover_refusal_score | #20 / 46 | ↓3.72 | Source ↗official | |
| Governance Decay under Passive Context Compactiongovernance_retention_score | #6 / 7 | ↑41 | Source ↗official | |
| HarmVideoBenchharmful_video_safety_recognition_reasoning | #14 / 19 | ↑0.712 | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #12 / 54 | ↑28.21 | Source ↗official | |
| LiveSecBenchethics | #7 / 43 | ↑79.3 | Source ↗official | |
| LiveSecBenchfactuality | #4 / 43 | ↑81.18 | Source ↗official | |
| LiveSecBenchlegality | #13 / 43 | ↑67.47 | Source ↗official | |
| LiveSecBenchprivacy | #17 / 43 | ↑60.89 | Source ↗official | |
| LiveSecBenchpsychological_health | #4 / 43 | ↑85.12 | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #37 / 50 | ↓93.5 | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #26 / 31 | ↓8.967 | Source ↗self run | |
| MASKlying_probability_pct | #21 / 53 | ↓29.53 | Source ↗official | |
| PHAREbias_resistance_diagnostic | #62 / 66 | ↑0.29 | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #18 / 70 | ↑0.8069 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #10 / 70 | ↑0.972 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #23 / 67 | ↑0.613 | Source ↗official | |
| SABERoverall_safety_rate | #8 / 13 | ↑23.9 | Source ↗official | |
| SABERscenario_a_safety_rate | #6 / 13 | ↑28.95 | Source ↗official | |
| SABERscenario_b_safety_rate | #9 / 13 | ↑28.24 | Source ↗official | |
| SABERscenario_c_safety_rate | #9 / 13 | ↑14.48 | Source ↗official | |
| SM-Benchadversarial | #22 / 78 | ↑84.39 | Source ↗official | |
| SM-Benchambiguous_interpretation | #18 / 78 | ↑89.29 | Source ↗official | |
| SM-Benchanti_hallucination | #38 / 78 | ↑93.19 | Source ↗official | |
| SM-Bencheq_boundaries | #55 / 78 | ↑58.15 | Source ↗official | |
| SM-Benchoverfit | #43 / 78 | ↑68.85 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #53 / 105 | ↑6.8 | Source ↗official | |
| SpeciEvalland_animal_4ns | #20 / 105 | ↓4.28 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #12 / 105 | ↓4.33 | Source ↗official | |
| SpeciEvalspeciesism | #29 / 105 | ↓1.73 | Source ↗official | |
| StereoTales Harmful Associationsbenign_significant_association_score | #15 / 23 | ↑85.53 | Source ↗official | |
| TACbase_welfare_rate | #45 / 74 | ↑24.36 | Source ↗self run | |
| ToolPrivacyBenchprivate_mt_poi | #8 / 9 | ↓27.74 | Source ↗official | |
| ToolPrivacyBenchpublic_mt_poi | #1 / 9 | ↓15.81 | Source ↗official | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #81 / 94 | ↑85.8 | Source ↗official |
Values evaluations
Descriptive values and political-framing results are separate from safety/ethics ranks. Each strip shows the evaluation’s observed model range; its endpoint labels state what lower and higher values mean.
UGI Political Values
Taiwan Sovereignty Benchmark Pro
| Dimension | Value | Distribution |
|---|---|---|
| Pro-Taiwan rubric compatibility | 15 | |
| Warning-phrase rate | 40 | |
| Soft-censorship rate | 0 | |
| API-error rate | 0 |
CCPBench political narrative alignment
| Dimension | Value | Distribution |
|---|---|---|
| CCP-narrative alignment — all questions | 2.86 | |
| CCP-narrative alignment — China topics | 3.34 | |
| CCP-narrative alignment — non-China controls | 1.41 |
Agent-ValueBench Moral Foundations (MFT08)
Agent-ValueBench HEXACO
| Dimension | Value | Distribution |
|---|---|---|
| Openness to experience | 5.8 | |
| Honesty-humility | 6 | |
| Extraversion | 6.7 | |
| Agreeableness | 6.8 | |
| Conscientiousness | 7.3 |