SAI LabsBook a demo

Safety IndexOpenAIGPT-5.6 Sol

GPT-5.6 Sol

OpenAIProprietaryReleased 2026-07Reasoning model

SAI Safety Index80provisionalHas 1 of the 4 pillars needed
Strongest pillar80Refusal calibration
Weakest pillar–No data
Coverage1of 5 pillars1 counted measurements, 0 by SAI

Summary

GPT-5.6 Sol has data in 1 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 80. Its strongest area is refusal calibration at 80.

1 measurements count toward the score, 0 of them from SAI's own runs. Not yet measured: adversarial robustness, agentic safety, interaction and honesty, people and society.

Pillar scores

How GPT-5.6 Sol 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 requests80 from 1 evaluation
  • Over-refusalnot measured

Rank 21 of 39 scored models.

Adversarial robustness

Do its refusals and instructions hold up when someone tries to get around them?

Not measured for GPT-5.6 Sol yet.
  • Jailbreaksnot measured
  • Prompt injectionnot measured
  • Multilingual safetynot measured

9 models scored.

Agentic safety

Does it behave safely when it has tools and autonomy?

Not measured for GPT-5.6 Sol yet.
  • 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?

Not measured for GPT-5.6 Sol yet.
  • Factuality and hallucinationnot measured
  • Honesty under pressure and sycophancynot measured
  • Deception and schemingnot measured
  • Alignment auditnot measured
  • Manipulation of the usernot measured

29 models scored.

People and society

Does it protect vulnerable people and the public interest?

Not measured for GPT-5.6 Sol yet.
  • 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.

EvaluationValueScoreSourceSetting
Refusal calibration
Harmful requests
Scale FORTRESS (risk score)average risk score 0-100 · Lower is better20.1%95% CI 18.3% to 22.0%80Scale AI Fortress leaderboard
Independent evaluator
Published by Scale AI Fortress leaderboard.
Full configurationFortress row 'gpt-5.6-sol'
Other evaluators' scores (not counted)
Vectara HHEM hallucination rate% · Lower is better12.4%context, not countedVectara 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 'GPT-5.6 Sol'