live
weekly refresh
basedagi.org
benchmark evidence

UGI — ugi-willingness

UGI Leaderboard — ugi-willingness score from DontPlanToEnd/UGI-Leaderboard

winner on UGI — ugi-willingness
direct benchmark result, not a broad vertical composite | source row dated 2025-09-30
scored on 2026-04-25 · stale source data (35d)
latest mapped results | top 20
#ModelScoreEvidenceTested
1Mistral Large
Mistralai
75.0
model-only
independent_benchmark
2025-09-30
2Mistral: Mistral Nemo
Mistralai
55.0
model-only
independent_benchmark
2025-09-06
3OpenAI: GPT-5.4
Openai
55.0
model-only
independent_benchmark
2026-03-05
4Qwen: Qwen3 30B A3B
Qwen
50.0
model-only
independent_benchmark
2025-11-08
5Meta: Llama 3.3 70B Instruct
Meta Llama
40.0
model-only
independent_benchmark
2025-09-10
6Qwen: Qwen3 32B
Qwen
38.0
model-only
independent_benchmark
2025-09-13
7Qwen: Qwen3 235B A22B
Qwen
32.0
model-only
independent_benchmark
2025-09-21
8OpenAI: GPT-5.5
Openai
30.0
model-only
independent_benchmark
2026-04-25
9DeepSeek: R1
Deepseek
28.0
model-only
independent_benchmark
2025-09-21
10OpenAI: GPT-5.1
Openai
28.0
model-only
independent_benchmark
2025-11-16
11OpenAI: o1
Openai
25.0
model-only
independent_benchmark
2025-09-24
12OpenAI: o4 Mini
Openai
22.0
model-only
independent_benchmark
2025-10-04
13OpenAI: GPT-4.1
Openai
22.0
model-only
independent_benchmark
2025-09-24
14Google: Gemini 2.5 Pro
Google
22.0
model-only
independent_benchmark
2025-09-09
15OpenAI: o3
Openai
22.0
model-only
independent_benchmark
2025-09-24
16OpenAI: GPT-5
Openai
22.0
model-only
independent_benchmark
2025-09-24
17OpenAI: GPT-5.2
Openai
22.0
model-only
independent_benchmark
2026-03-04
18Google: Gemma 3 27B
Google
20.0
model-only
independent_benchmark
2025-09-06
19MoonshotAI: Kimi K2 Thinking
Moonshotai
18.0
model-only
independent_benchmark
2025-11-07
20Anthropic: Claude Sonnet 4
Anthropic
12.0
model-only
independent_benchmark
2025-09-12
what this result means

UGI Leaderboard — ugi-willingness score from DontPlanToEnd/UGI-Leaderboard

This benchmark contributes direct public evidence. Read its scope before generalizing the result.

A win here is a win on UGI — ugi-willingness. Broad task pages require independent corroboration before naming a general winner.

source record
category: uncensored
metric: accuracy
matched models: 20
latest source date: 2026-04-25
direction: higher is better
inspect upstream source ->