Benchmark Almanac

LLM Sycophancy Benchmark

Grok 4.5 0% leads since July 2026
Not in model cards

Models judge the same dispute from opposite narrators’ perspectives. Contradictory judgments count toward the sycophancy rate.

The author publishes the model comparison in the repository.

Made by
Lech Mazur
Kind
Deception and alignment
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Propensity
Language
English
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Claude Fable 5 #12 on the capability index
Last model added
July 2026
Scores refreshed
November 20th, 2026
24
Entries on its own leaderboard, 20 models we track

2 models share the top score of 0% among models released by July 2026 2 models share the top score of 0% among models released by July 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in July 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Grok 4.5 high reasoning 0%
  2. 2 GPT-5.6 Terra high reasoning 0%
  3. 3 Claude Fable 5 medium reasoning 0.50%
  4. 4 Tencent Hy3 high 0.50%
  5. 5 Gemini 3.6 Flash 0.50%
  6. 6 GPT-5.6 Luna high reasoning 1.00%
  7. 7 Gemini 3.5 Flash-Lite 1.00%
  8. 8 Qwen3.7-Max 2026-06-08 1.50%
  9. 9 Baidu Ernie 5.1 2.00%
  10. 10 Qwen3.7 Flash 2.50%
  11. 11 Xiaomi MiMo V2.5 Pro 2.60%
  12. 12 GPT-5.6 Sol high reasoning 3.00%
  13. 13 Inkling high reasoning 3.50%
  14. 14 MiniMax-M3 3.60%
  15. 15 Kimi K3 4.50%
  16. 16 Gemma 4 31B IT reasoning 4.60%
  17. 17 DeepSeek-V4-Pro 5.10%
  18. 18 Qwen3.7-Plus 5.10%
  19. 19 DeepSeek-V4-Flash high reasoning 5.60%
  20. 20 Claude Sonnet 5 high reasoning 9.10%
  21. 21 GLM-5.2 max reasoning 12.6%
  22. 22 ByteDance Seed2.1 Pro 14.6%
  23. 23 Arcee Trinity Large Thinking 18.9%
  24. 24 Mistral Medium 3.5 high 22.4%

Source: Official LLM Sycophancy Benchmark leaderboard

#EntryScoreSettingDateSource
1 Grok 4.5 0% high reasoning Jul 2026 Official leaderboard
2 GPT-5.6 Terra 0% high reasoning Jul 2026 Official leaderboard
3 Claude Fable 5 0.50% medium reasoning Jun 2026 Official leaderboard
4 Tencent Hy3 0.50% high Jul 2026 Official leaderboard
5 Gemini 3.6 Flash 0.50% Jul 2026 Official leaderboard
6 GPT-5.6 Luna 1.00% high reasoning Jul 2026 Official leaderboard
7 Gemini 3.5 Flash-Lite 1.00% Jul 2026 Official leaderboard
8 Qwen3.7-Max 1.50% 2026-06-08 May 2026 Official leaderboard
9 Baidu Ernie 5.1 2.00% Official leaderboard
10 Qwen3.7 Flash 2.50% Jul 2026 Official leaderboard
11 Xiaomi MiMo V2.5 Pro 2.60% Official leaderboard
12 GPT-5.6 Sol 3.00% high reasoning Jul 2026 Official leaderboard
13 Inkling 3.50% high reasoning Jul 2026 Official leaderboard
14 MiniMax-M3 3.60% Jun 2026 Official leaderboard
15 Kimi K3 4.50% Jul 2026 Official leaderboard
16 Gemma 4 31B IT 4.60% reasoning Apr 2026 Official leaderboard
17 DeepSeek-V4-Pro 5.10% Apr 2026 Official leaderboard
18 Qwen3.7-Plus 5.10% Jun 2026 Official leaderboard
19 DeepSeek-V4-Flash 5.60% high reasoning Apr 2026 Official leaderboard
20 Claude Sonnet 5 9.10% high reasoning Jun 2026 Official leaderboard
21 GLM-5.2 12.6% max reasoning Jun 2026 Official leaderboard
22 ByteDance Seed2.1 Pro 14.6% Official leaderboard
23 Arcee Trinity Large Thinking 18.9% Official leaderboard
24 Mistral Medium 3.5 22.4% high Apr 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Grok 4.5 0% high reasoning Jul 2026 Official leaderboard
2 GPT-5.6 Terra 0% high reasoning Jul 2026 Official leaderboard
3 Claude Fable 5 0.50% medium reasoning Jun 2026 Official leaderboard
4 Tencent Hy3 0.50% high Jul 2026 Official leaderboard
5 Gemini 3.6 Flash 0.50% Jul 2026 Official leaderboard
6 GPT-5.6 Luna 1.00% high reasoning Jul 2026 Official leaderboard
7 Gemini 3.5 Flash-Lite 1.00% Jul 2026 Official leaderboard
8 Qwen3.7-Max 1.50% 2026-06-08 May 2026 Official leaderboard
9 Qwen3.7 Flash 2.50% Jul 2026 Official leaderboard
10 GPT-5.6 Sol 3.00% high reasoning Jul 2026 Official leaderboard
11 Inkling 3.50% high reasoning Jul 2026 Official leaderboard
12 MiniMax-M3 3.60% Jun 2026 Official leaderboard
13 Kimi K3 4.50% Jul 2026 Official leaderboard
14 Gemma 4 31B IT 4.60% reasoning Apr 2026 Official leaderboard
15 DeepSeek-V4-Pro 5.10% Apr 2026 Official leaderboard
16 Qwen3.7-Plus 5.10% Jun 2026 Official leaderboard
17 DeepSeek-V4-Flash 5.60% high reasoning Apr 2026 Official leaderboard
18 Claude Sonnet 5 9.10% high reasoning Jun 2026 Official leaderboard
19 GLM-5.2 12.6% max reasoning Jun 2026 Official leaderboard
20 Mistral Medium 3.5 22.4% high Apr 2026 Official leaderboard

4 models have held number one on the official leaderboard since 2023

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind Qwen Anthropic xAI
Own leaderboard Not tested since Jul 2026

Source: Official LLM Sycophancy Benchmark leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Grok 4.5 Qwen3.7-Max 0% 20% 40% 60% 80% 100% 2026 Grok 4.5

Source: Official LLM Sycophancy Benchmark leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Safety and alignment

All 636 in Safety and alignment
  • FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
  • StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is. Inkling 98.6%
  • Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts. Gemini 4 Carbon 0%
  • Covert misbehavior (Thinking Machines) In use Agentic tasks that check whether a model misbehaves and conceals it. Scored by the rate of covert misbehavior. Inkling 2 Mini 8.30%
  • Destructive actions (Thinking Machines) In use Agentic tasks that check whether a model takes destructive actions it was not authorized to take. Scored by the rate of destructive actions. Inkling 2 Small 4.70%
  • Misbehavior (Thinking Machines) In use Agentic tasks that check whether a model fabricates results, ignores instructions or hides what it did when a task is hard. Scored by the rate of misbehavior. Inkling 2 Mini 9.80%

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (24 entries).
  • Pass Models scored 20 models we track, plus 4 other entries on its leaderboard.
  • Pass Newest score The newest is from July 2026, for Qwen3.7 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Fable 5 (#12).
  • Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 20 rows for models we track and 4 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check