Model profile
GPT 5.6 Terra
OpenAIdeveloper
2026-07-09release date
#7 / 267overall rank
17eval lineages
1discovery sources
Evidence summary
GPT 5.6 Terra has an estimated overall rank of #7; its 90% source-sensitivity interval is #5–#60. Published evidence spans 17 evals and 7 of 7 behavior components. Its strongest relative result is SpeciEval (speciesism, #5 of 102); its weakest is GPT-5.6 system card — disallowed content with challenging prompts (gore_not_unsafe, #6 of 7).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗gpt-5-6-terra
- OpenRouter ↗openai/gpt-5.6-terra
- Official model documentation ↗Family-level model document · openai · first party
- 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 | Better | Source |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #202 / 311 | 0.8519 | ↓ lower | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #23 / 105 | 0.495 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #11 / 43 | 54.6 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #22 / 48 | 66.2 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #20 / 49 | 50.7 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #3 / 45 | 80.4 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #6 / 51 | 6.7 | ↓ lower | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #14 / 32 | 46.2 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #41 / 48 | 22.2 | ↓ lower | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #35 / 260 | 39.28 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #217 / 260 | 80 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #24 / 260 | 97.78 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #198 / 258 | 92.36 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — disallowed content with challenging promptsextremism_not_unsafe | #4 / 7 | 0.981 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — disallowed content with challenging promptsgore_not_unsafe | #6 / 7 | 0.6 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — disallowed content with challenging promptshate_not_unsafe | #1 / 7 | 1 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — disallowed content with challenging promptsnonviolent_illicit_not_unsafe | #4 / 7 | 0.99 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — disallowed content with challenging promptsself_harm_not_unsafe | #2 / 7 | 0.962 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — disallowed content with challenging promptssexual_minors_not_unsafe | #1 / 7 | 0.974 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — disallowed content with challenging promptssexual_not_unsafe | #1 / 7 | 0.966 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — disallowed content with challenging promptsviolent_illicit_not_unsafe | #4 / 7 | 0.952 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — first-person fairnessharm_overall_pct | #2 / 7 | 0.88 | ↓ lower | Source ↗official | |
| GPT-5.6 system card — prompt-injection robustnessconnectors_injection_resistance | #1 / 7 | 1 | ↑ higher | Source ↗official | |
| GPT-5.6 system card — prompt-injection robustnesssearch_function_calling_injection_resistance | #1 / 6 | 0.946 | ↑ higher | Source ↗official | |
| Gray Swan indirect prompt injection (15 attempts)attack_success_probability_k15_pct | #9 / 13 | 30.4 | ↓ lower | Source ↗official | |
| HUMAINE Trust, Ethics and Safetytrust_ethics_safety_score | #19 / 54 | 27.66 | ↑ higher | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #3 / 50 | 80.4 | ↓ lower | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #18 / 31 | 8.7 | ↓ lower | Source ↗official | |
| Manager Coercion Benchfabrication_rate | #1 / 13 | 0 | ↓ lower | Source ↗official | |
| SM-Benchadversarial | #26 / 73 | 83.66 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #40 / 73 | 83.63 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #41 / 73 | 90.84 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #31 / 73 | 66.01 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #55 / 73 | 53.55 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #27 / 102 | 6.925 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #22 / 102 | 4.29 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #5 / 102 | 4.225 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #5 / 102 | 1.265 | ↓ lower | Source ↗official | |
| TACbase_welfare_rate | #29 / 68 | 29.49 | ↑ higher | 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.