# Spiral-Bench v1.0 > Source: Benchize (https://benchize.com) — https://benchize.com/benchmarks/eq-bench%3Aspiral-bench-v1.0. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/benchmarks/eq-bench%3Aspiral-bench-v1.0. The original Spiral-Bench protocol for simulated conversations with users vulnerable to delusional thinking. A single judge grades protective and risky behaviors; these legacy scores remain separate from the revised protocol. Canonical URL: https://benchize.com/benchmarks/eq-bench%3Aspiral-bench-v1.0 Last updated: Wed, 21 Oct 2026 18:52:20 GMT Reproduction and quotation must name Benchize, link to the canonical HTML URL, and retain individual source citations. The original Spiral-Bench protocol for simulated conversations with users vulnerable to delusional thinking. A single judge grades protective and risky behaviors; these legacy scores remain separate from the revised protocol. Safety & robustness · Models tested: 11 · Score publishers: [eq bench](https://eqbench.com/spiral-bench_v1.0.html) ## Model results | # | Model | Score | ± | | | --- | --- | --- | --- | --- | | 1 | [O3](https://benchize.com/models/o3.md) | 86.1% | CI 86.1–86.1% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 86.1 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion o3 ### Reported statistics Uncertainty Confidence interval: 86.1–86.1% confidenceIntervalHigh 86.1 confidenceIntervalLow 86.1 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label o3 Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/12` | | | | | | 2 | [gpt-oss-120b](https://benchize.com/models/gpt-oss-120b.md) | 81.4% | CI 81.4–81.4% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 81.4 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion gpt-oss-120b ### Reported statistics Uncertainty Confidence interval: 81.4–81.4% confidenceIntervalHigh 81.4 confidenceIntervalLow 81.4 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label gpt-oss-120b Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/8` | | | | | | 3 | [Claude Sonnet 4.5](https://benchize.com/models/claude-sonnet-4-5.md) | 76.4% | CI 76.4–76.4% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 76.4 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion claude-sonnet-4.5 ### Reported statistics Uncertainty Confidence interval: 76.4–76.4% confidenceIntervalHigh 76.4 confidenceIntervalLow 76.4 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label claude-sonnet-4.5 Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/13` | | | | | | 4 | [O4 Mini](https://benchize.com/models/o4-mini.md) | 73.3% | CI 73.3–73.3% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 73.3 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion o4-mini ### Reported statistics Uncertainty Confidence interval: 73.3–73.3% confidenceIntervalHigh 73.3 confidenceIntervalLow 73.3 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label o4-mini Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/11` | | | | | | 5 | [Kimi K2](https://benchize.com/models/kimi-k2.md) | 73% | CI 73–73% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 73.0 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion kimi-k2 ### Reported statistics Uncertainty Confidence interval: 73–73% confidenceIntervalHigh 73.0 confidenceIntervalLow 73.0 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label kimi-k2 Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/9` | | | | | | 6 | [Gemini 2.5 Flash](https://benchize.com/models/gemini-2-5-flash.md) | 49.1% | CI 49.1–49.1% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 49.1 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion gemini-2.5-flash ### Reported statistics Uncertainty Confidence interval: 49.1–49.1% confidenceIntervalHigh 49.1 confidenceIntervalLow 49.1 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label gemini-2.5-flash Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/4` | | | | | | 7 | [Llama 4 Maverick](https://benchize.com/models/llama-4-maverick.md) | 48.1% | CI 48.1–48.1% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 48.1 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion llama-4-maverick ### Reported statistics Uncertainty Confidence interval: 48.1–48.1% confidenceIntervalHigh 48.1 confidenceIntervalLow 48.1 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label llama-4-maverick Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/10` | | | | | | 8 | [Claude 3.5 Sonnet](https://benchize.com/models/claude-3-5-sonnet.md) | 47.7% | CI 47.7–47.7% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 47.7 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion claude-3.5-sonnet ### Reported statistics Uncertainty Confidence interval: 47.7–47.7% confidenceIntervalHigh 47.7 confidenceIntervalLow 47.7 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label claude-3.5-sonnet Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/1` | | | | | | 9 | [Gemini 2.5 Pro](https://benchize.com/models/gemini-2-5-pro.md) | 43.5% | CI 43.5–43.5% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 43.5 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion gemini-2.5-pro ### Reported statistics Uncertainty Confidence interval: 43.5–43.5% confidenceIntervalHigh 43.5 confidenceIntervalLow 43.5 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label gemini-2.5-pro Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/5` | | | | | | 10 | [Claude Sonnet 4](https://benchize.com/models/claude-sonnet-4.md) | 41.1% | CI 41.1–41.1% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 41.1 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion claude-sonnet-4 ### Reported statistics Uncertainty Confidence interval: 41.1–41.1% confidenceIntervalHigh 41.1 confidenceIntervalLow 41.1 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label claude-sonnet-4 Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/2` | | | | | | 11 | [R1 0528](https://benchize.com/models/deepseek-r1-0528.md) | 22.4% | CI 22.4–22.4% | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score_norm` \| 22.4 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol spiral-bench-v1.0 sourceModelVersion deepseek-r1-0528 ### Reported statistics Uncertainty Confidence interval: 22.4–22.4% confidenceIntervalHigh 22.4 confidenceIntervalLow 22.4 ### Source [Original source](https://eqbench.com/spiral-bench_v1.0.js) Source eq bench Original model label deepseek-r1-0528 Source version spiral-bench-v1.0 Observed 2026-10-21T16:35:04.619Z Evidence SHA-256 `d939ed9ba6d42226765fb4acf3e440127472566e1b5c2894d4ca25baa5ae75d2` Source locator `/leaderboardDataDelusion/3` | | | | | --- Cite as: Spiral-Bench v1.0 — Benchize, https://benchize.com/benchmarks/eq-bench%3Aspiral-bench-v1.0 (retrieved 2026-10-21)