Model profile
Inkling
Evidence summary
Safety. Inkling has an estimated Safety rank of #49; its 90% source-sensitivity interval is #13–#210. Its behavior-only rank is #43; company governance moves the combined estimate to #49. Published Safety evidence spans 9 eval lineages and 6 of 7 components. Its strongest relative result is SM-Bench (ambiguous_interpretation, #2 of 84); its weakest is SM-Bench (eq_boundaries, #79 of 84).
Freedom. Inkling has an estimated Freedom rank of #569; its 90% source-sensitivity interval is #308–#643. Published Freedom evidence spans 3 eval lineages and 1 of 1 components. Its strongest relative result is Inkling-Small model card — StrongREJECT (safety_rate, #4 of 10); its weakest is SM-Bench (eq_boundaries, #79 of 84).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗inkling
- Hugging Face ↗thinkingmachines/Inkling
- OpenRouter ↗thinkingmachines/inkling
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
- thinkingmachines.ai ↗
Safety evals
Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #119 / 345 | ↓0.6766 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #34 / 248 | ↑39.02 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #58 / 248 | ↑91.33 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #22 / 248 | ↑97.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #73 / 246 | ↑98 | Source ↗official | |
| Inkling-Small model card — FORTRESSbenign_answer_rate | #4 / 10 | ↑95.9 | Source ↗official | |
| Inkling-Small model card — FORTRESSharmful_refusal_rate | #4 / 10 | ↑78 | Source ↗official | |
| Inkling-Small model card — StrongREJECTsafety_rate | #6 / 10 | ↑98.6 | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #31 / 37 | ↓8.933 | Source ↗self run | |
| Opposite-Narrator Sycophancysycophancy_rate_pct | #13 / 24 | ↓3.5 | Source ↗official | |
| Pander Scoreconversational_absolute_pander_score | #19 / 26 | ↓18.37 | Source ↗official | |
| Pander Scoreinstructional_absolute_pander_score | #17 / 26 | ↓45.4 | Source ↗official | |
| SM-Benchadversarial | #40 / 84 | ↑82.44 | Source ↗official | |
| SM-Benchambiguous_interpretation | #2 / 84 | ↑96.73 | Source ↗official | |
| SM-Benchanti_hallucination | #69 / 84 | ↑83.25 | Source ↗official | |
| SM-Bencheq_boundaries | #79 / 84 | ↑48.03 | Source ↗official | |
| SM-Benchoverfit | #45 / 84 | ↑72.68 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #25 / 123 | ↑6.97 | Source ↗official | |
| SpeciEvalland_animal_4ns | #8 / 123 | ↓3.92 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #8 / 123 | ↓4.23 | Source ↗official | |
| SpeciEvalspeciesism | #16 / 123 | ↓1.48 | Source ↗official | |
| TACbase_welfare_rate | #49 / 87 | ↑25.64 | Source ↗self run |
Freedom evals
Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #188 / 248 | ↓91.33 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #223 / 248 | ↓97.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #172 / 246 | ↓98 | Source ↗official | |
| Inkling-Small model card — FORTRESSbenign_answer_rate | #4 / 10 | ↑95.9 | Source ↗official | |
| Inkling-Small model card — FORTRESSharmful_refusal_rate | #7 / 10 | ↓78 | Source ↗official | |
| Inkling-Small model card — StrongREJECTsafety_rate | #4 / 10 | ↓98.6 | Source ↗official | |
| SM-Benchadversarial | #43 / 84 | ↓82.44 | Source ↗official | |
| SM-Bencheq_boundaries | #79 / 84 | ↑48.03 | Source ↗official | |
| SM-Benchoverfit | #45 / 84 | ↑72.68 | Source ↗official |