live
weekly refresh
basedagi.org
benchmark evidence

UGI — ugi-natural-intelligence

UGI Leaderboard — ugi-natural-intelligence score from DontPlanToEnd/UGI-Leaderboard

winner on UGI — ugi-natural-intelligence
direct benchmark result, not a broad vertical composite | source row dated 2026-04-25
scored on 2026-04-25 · stale source data (35d)
latest mapped results | top 20
#ModelScoreEvidenceTested
1OpenAI: GPT-5.5
Openai
79.3
model-only
independent_benchmark
2026-04-25
2OpenAI: GPT-5.4
Openai
73.5
model-only
independent_benchmark
2026-03-05
3OpenAI: GPT-5
Openai
69.3
model-only
independent_benchmark
2025-09-24
4OpenAI: o3
Openai
66.2
model-only
independent_benchmark
2025-09-24
5OpenAI: o1
Openai
64.2
model-only
independent_benchmark
2025-09-24
6Google: Gemini 2.5 Pro
Google
64.0
model-only
independent_benchmark
2025-09-09
7OpenAI: GPT-5.1
Openai
62.4
model-only
independent_benchmark
2025-11-16
8OpenAI: GPT-5.2
Openai
61.5
model-only
independent_benchmark
2026-03-04
9OpenAI: GPT-4.1
Openai
59.9
model-only
independent_benchmark
2025-09-24
10MoonshotAI: Kimi K2 Thinking
Moonshotai
54.4
model-only
independent_benchmark
2025-11-07
11DeepSeek: R1
Deepseek
52.0
model-only
independent_benchmark
2025-09-21
12Anthropic: Claude Sonnet 4
Anthropic
51.1
model-only
independent_benchmark
2025-09-12
13OpenAI: o4 Mini
Openai
45.5
model-only
independent_benchmark
2025-10-04
14Mistral Large
Mistralai
37.6
model-only
independent_benchmark
2025-09-30
15Qwen: Qwen3 235B A22B
Qwen
36.5
model-only
independent_benchmark
2025-09-21
16Google: Gemma 3 27B
Google
34.5
model-only
independent_benchmark
2025-09-06
17Meta: Llama 3.3 70B Instruct
Meta Llama
31.4
model-only
independent_benchmark
2025-09-10
18Mistral: Mistral Nemo
Mistralai
20.8
model-only
independent_benchmark
2025-09-06
19Qwen: Qwen3 32B
Qwen
20.1
model-only
independent_benchmark
2025-09-13
20Qwen: Qwen3 30B A3B
Qwen
19.2
model-only
independent_benchmark
2025-11-08
what this result means

UGI Leaderboard — ugi-natural-intelligence 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-natural-intelligence. 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 ->