Safety IndexMoonshot AIKimi K2.5
Kimi K2.5
Moonshot AIOpen weightsReleased 2026-01Reasoning model
Summary
Kimi K2.5 has data in 2 of the five pillars, short of the 4 needed for a headline number, so it is listed as partly measured; the mean of its measured pillars is 65. Its strongest area is interaction and honesty at 71, and its weakest is refusal calibration at 59.
2 measurements count toward the score, 0 of them from SAI's own runs. Not yet measured: adversarial robustness, agentic safety, people and society.
Pillar scores
How Kimi K2.5 compares with every other model scored in the pillar, and which sub-categories the score rests on.
Refusal calibration
Does the model refuse what it should and help with what it may?
- Harmful requests59 from 1 evaluation
- Over-refusalnot measured
Rank 33 of 39 scored models.
Adversarial robustness
Do its refusals and instructions hold up when someone tries to get around them?
- Jailbreaksnot measured
- Prompt injectionnot measured
- Multilingual safetynot measured
9 models scored.
Agentic safety
Does it behave safely when it has tools and autonomy?
- Agent misusenot measured
- Reward hacking and task integritynot measured
- Overreach and destructive actionsnot measured
3 models scored.
Interaction and honesty
Is it honest with the person it is talking to?
- Factuality and hallucinationnot measured
- Honesty under pressure and sycophancy71 from 1 evaluation
- Deception and schemingnot measured
- Alignment auditnot measured
- Manipulation of the usernot measured
Rank 13 of 29 scored models.
People and society
Does it protect vulnerable people and the public interest?
- Self-harm and wellbeingnot measured
- Child safetynot measured
- Bias, fairness and neutralitynot measured
- Privacy: memorisation and leakagenot measured
- Privacy: surveillance assistancenot measured
- Elections and influence operationsnot measured
3 models scored.
Every measurement
These are the raw numbers behind the scores, with who measured them, when, and how. Rows marked counted feed the index. The others are shown for context, such as benign twins of harm rows, retired evaluations that every frontier model passes, tests of dangerous knowledge, other evaluators' overall scores, and the developer's own figures.
| Evaluation | Value | Score | Source | Setting |
|---|---|---|---|---|
| Refusal calibration | ||||
| Harmful requests | ||||
| Scale FORTRESS (risk score) | 41.1%95% CI 37.5% to 44.7% | 59 | Scale AI Fortress leaderboard Independent evaluator | Published by Scale AI Fortress leaderboard.Full configurationFortress row 'kimi-k2.5' |
| Interaction and honesty | ||||
| Honesty under pressure and sycophancy | ||||
| MASK | 70.5%95% CI 66.0% to 75.0%n = 500 | 71 | Scale AI SEAL / CAIS MASK leaderboard Independent evaluator | Published by Scale AI SEAL / CAIS MASK leaderboard.Full configurationScale leaderboard on the private 500-example split, not the public split; MASK private split, 500 examples; honesty score = 1 - P(lie); row 'kimi-k2.5' |
| Other evaluators' scores (not counted) | ||||
| Phare average safety | 67.0% | context, not counted | Giskard Phare leaderboard Independent evaluator · 2026-07-09 | Published by Giskard Phare leaderboard (2026-07-09).Full configurationPhare row 'Kimi K2.5' |
| Phare bias resistance | 29% | context, not counted | Giskard Phare leaderboard Independent evaluator · 2026-07-09 | Published by Giskard Phare leaderboard (2026-07-09).Full configurationPhare row 'Kimi K2.5' |
| Phare hallucination resistance | 80.7% | context, not counted | Giskard Phare leaderboard Independent evaluator · 2026-07-09 | Published by Giskard Phare leaderboard (2026-07-09).Full configurationPhare row 'Kimi K2.5' |
| Phare harmfulness resistance | 97.2% | context, not counted | Giskard Phare leaderboard Independent evaluator · 2026-07-09 | Published by Giskard Phare leaderboard (2026-07-09).Full configurationPhare row 'Kimi K2.5' |
| Phare jailbreak resistance | 61.3% | context, not counted | Giskard Phare leaderboard Independent evaluator · 2026-07-09 | Published by Giskard Phare leaderboard (2026-07-09).Full configurationPhare row 'Kimi K2.5' |
| Vectara HHEM hallucination rate | 14.2% | context, not counted | Vectara Hallucination Leaderboard (HHEM-2.3) Independent evaluator · 2026-10-22 | Published by Vectara Hallucination Leaderboard (HHEM-2.3) (2026-10-22).Full configurationsummarization of ~1,000 short documents; row 'Kimi K2.5' |