UGI Leaderboard

Name: UGI Leaderboard results
Creator: UGI Leaderboard
License: https://www.apache.org/licenses/LICENSE-2.0

How closely the model matches the writing style of a given example.

Last updated 2 Oct 2026

Results dated: 6 Sep 2025 to 2 Oct 2026
Results: 379 configurations of 230 models
Unit: score from 0 to 1
Licence: Apache License 2.0

Style adherence: Trinity Large Preview

Top 15 of 230 results · score from 0 to 1, higher is better · ≈ cannot be told apart from the leader. Choose a model to highlight it.Clear highlight

1≈ Claude Opus 5.5 (max reasoning)Anthropic 0.43
1≈ Claude Fable 5.1 (high reasoning)Anthropic 0.43
3 Claude Opus 4.6 (high reasoning)Anthropic 0.42
4 Claude Opus 4.7 (high reasoning)Anthropic 0.41
4 Claude Sonnet 4.6 (max reasoning)Anthropic 0.41
4 GPT-5.5 (high reasoning)OpenAI 0.41
4 MiniMax-M2.1MiniMax 0.41
4 Claude Sonnet 4.5 (reasoning on)Anthropic 0.41
4 Claude Fable 5 (extra-high reasoning)Anthropic 0.41
10 Claude Opus 4.8 (max reasoning)Anthropic 0.40
10 MiMo-V2.5-Pro (reasoning on)Xiaomi 0.40
10 MiMo-V2.5 (reasoning on)Xiaomi 0.40
10 GLM-5 (no reasoning)Z.ai 0.40
10 Trinity Large PreviewArcee AI 0.40
10 GPT-5.4 (high reasoning)OpenAI 0.40

Full results

UGI: style adherence, score from 0 to 1, higher is better
#	Model	Style adherence score from 0 to 1, higher is better	Price $ per million tokens, in / out
1≈	Claude Opus 5.5 (max reasoning)Anthropic · best of 5 settings	0.43	$4 / $20
1≈	Claude Fable 5.1 (high reasoning)Anthropic · best of 3 settings	0.43	$10 / $50
3	Claude Opus 4.6 (high reasoning)Anthropic · best of 4 settings	0.42	$5 / $25
4	Claude Opus 4.7 (high reasoning)Anthropic · best of 4 settings	0.41	$5 / $25
4	Claude Sonnet 4.6 (max reasoning)Anthropic · best of 4 settings	0.41	$3 / $15
4	GPT-5.5 (high reasoning)OpenAI · best of 5 settings	0.41	$5 / $30
4	MiniMax-M2.1MiniMax	0.41	$0.30 / $1.20
4	Claude Sonnet 4.5 (reasoning on)Anthropic · best of 2 settings	0.41	$3 / $15
4	Claude Fable 5 (extra-high reasoning)Anthropic · best of 5 settings	0.41	$10 / $50
10	Claude Opus 4.8 (max reasoning)Anthropic · best of 5 settings	0.40	$5 / $25
10	MiMo-V2.5-Pro (reasoning on)Xiaomi · best of 2 settings	0.40	$0.43 / $0.87
10	MiMo-V2.5 (reasoning on)Xiaomi · best of 2 settings	0.40	$0.17 / $0.34
10	GLM-5 (no reasoning)Z.ai · best of 2 settings	0.40	$0.95 / $2.55
10	Trinity Large PreviewArcee AI	0.40	–
10	GPT-5.4 (high reasoning)OpenAI · best of 5 settings	0.40	$2.50 / $15
10	Gemini 2.5 ProGoogle	0.40	$1.25 / $10
10	GPT-5.2 (low reasoning)OpenAI · best of 4 settings	0.40	$1.75 / $14
18	GPT-5.6 Terra (extra-high reasoning)OpenAI · best of 5 settings	0.39	$2 / $12
18	GPT-6 Sol (extra-high reasoning)OpenAI · best of 4 settings	0.39	$2 / $10
18	Claude Opus 4.5 (reasoning on)Anthropic · best of 2 settings	0.39	$5 / $25
18	GPT-5 (high reasoning)OpenAI · best of 2 settings	0.39	$1.25 / $10
18	Claude Haiku 4.5 (reasoning on)Anthropic · best of 2 settings	0.39	$1 / $5
18	GLM-4.6V (no reasoning)Z.ai · best of 2 settings	0.39	$0.30 / $0.90
18	Mistral Medium 3.1Mistral AI	0.39	$0.40 / $2
25	Kimi K2 ThinkingMoonshot AI	0.38	$0.60 / $2.50
25	MiniMax-M2.7MiniMax	0.38	$0.30 / $1.20
25	GPT-5.1 (low reasoning)OpenAI · best of 3 settings	0.38	$1.25 / $10
25	GLM-5.1 (reasoning on)Z.ai · best of 2 settings	0.38	$1.38 / $4.40
25	Qwen3-1.7B (no reasoning)Alibaba · best of 2 settings	0.38	–
25	GPT-5.6 Luna (extra-high reasoning)OpenAI · best of 5 settings	0.38	$0.20 / $1.20
25	Claude Sonnet 5 (extra-high reasoning)Anthropic · best of 5 settings	0.38	$2 / $10
25	Llama 3.1 70B InstructMeta	0.38	$0.40 / $0.40
25	GLM-4.6 (reasoning on)Z.ai · best of 2 settings	0.38	$0.50 / $2
25	GLM-4.7 (reasoning on)Z.ai · best of 2 settings	0.38	$0.54 / $1.98
25	MiniMax-M2.5MiniMax	0.38	$0.30 / $1.20
25	Mistral Large 3 (no reasoning)Mistral AI · best of 2 settings	0.38	$0.50 / $1.50
25	GPT-6 Astra (medium reasoning)OpenAI · best of 4 settings	0.38	$10 / $50
25	Claude Opus 4.1 (reasoning on)Anthropic · best of 2 settings	0.38	$15 / $75
39	GPT-6.1 Sol (high reasoning)OpenAI · best of 4 settings	0.37	$2 / $10
39	Step 3.5 FlashStepFun	0.37	$0.10 / $0.30
39	GLM-5.2 (no reasoning)Z.ai · best of 2 settings	0.37	$1.40 / $4.40
39	DeepSeek-V3DeepSeek	0.37	$0.26 / $1.03
39	DeepSeek-V3.1-Terminus (reasoning on)DeepSeek · best of 2 settings	0.37	$0.27 / $1
39	GPT-5.6 Sol (extra-high reasoning)OpenAI · best of 5 settings	0.37	$4 / $20
39	Qwen3-4B (no reasoning)Alibaba · best of 2 settings	0.37	–
39	Llama 4 MaverickMeta	0.37	$0.27 / $0.85
39	Muse Glimmer 30B (medium reasoning)Meta · best of 4 settings	0.37	$0.30 / $1.20
39	GLM-4.7-Flash (reasoning by prefill)Z.ai · best of 2 settings	0.37	$0.06 / $0.40
39	Qwen3-14B (no reasoning)Alibaba · best of 2 settings	0.37	$0.12 / $0.24
39	Gemini 3 Pro Preview (high reasoning)Google · best of 2 settings	0.37	–
39	Llama 3.3 70B InstructMeta	0.37	$0.59 / $0.79
39	Gemini 3 Flash Preview (high reasoning)Google · best of 3 settings	0.37	$0.50 / $3
39	DeepSeek-V3.2-SpecialeDeepSeek	0.37	–
39	Grok 4.7xAI	0.37	$2 / $6
39	Jamba Mini 1.7AI21 Labs	0.37	–
39	DeepSeek-V3.2 (reasoning on)DeepSeek · best of 2 settings	0.37	$0.30 / $0.96
39	DeepSeek-V4-Flash (0423, reasoning on)DeepSeek · best of 2 settings	0.37	$0.14 / $0.28
39	MiniMax-M2MiniMax	0.37	$0.30 / $1.20
39	Mistral Large 2.1 (2411)Mistral AI	0.37	–
60	Qwen3-8B (no reasoning)Alibaba · best of 2 settings	0.36	$0.12 / $0.46
60	Claude Opus 4 (reasoning on)Anthropic · best of 2 settings	0.36	–
60	Llama 3.1 405B InstructMeta	0.36	–
60	Mistral Large 2 (2407)Mistral AI	0.36	$2 / $6
60	Qwen2.5-72B-InstructAlibaba	0.36	$0.36 / $0.40
60	Qwen3-235B-A22B-Instruct-2507Alibaba	0.36	$0.15 / $0.75
60	Qwen3-30B-A3B (no reasoning)Alibaba · best of 2 settings	0.36	$0.12 / $0.50
60	DeepSeek-V3.1 (reasoning on)DeepSeek · best of 2 settings	0.36	$0.55 / $1.65
60	Llama 4 ScoutMeta	0.36	$0.18 / $0.59
60	GPT-5.3 ChatOpenAI	0.36	–
60	o1 (high reasoning)OpenAI · best of 2 settings	0.36	$15 / $60
60	Grok 3xAI	0.36	–
60	MiMo-V2-Flash (no reasoning)Xiaomi · best of 2 settings	0.36	–
60	Gemma 4 31B (reasoning by prefill)Google · best of 2 settings	0.36	$0.14 / $0.40
60	Falcon-H1 0.5B InstructTII	0.36	–
60	GLM-4.5 (reasoning on)Z.ai · best of 2 settings	0.36	$0.60 / $2.20
60	Jamba Large 1.7AI21 Labs	0.36	–
60	Claude Sonnet 4 (no reasoning)Anthropic · best of 2 settings	0.36	$3 / $15
60	DeepSeek-V4-Pro (0423, reasoning on)DeepSeek · best of 2 settings	0.36	$1.42 / $2.83
60	Rnj-1 InstructEssential AI	0.36	–
60	Mixtral 8x7B Instruct v0.1Mistral AI	0.36	–
60	Grok 4.6xAI	0.36	$2 / $6
60	Llama 3.2 3B InstructMeta	0.36	$0.05 / $0.33
60	Qwen3.5-122B-A10B (no reasoning)Alibaba · best of 2 settings	0.36	$0.26 / $2.08
60	Grok 4 (0709)xAI	0.36	–
60	Grok 4.20 Multi-Agent Beta (0309, 4 agents)xAI	0.36	–
60	Qwen3-VL-235B-A22B-InstructAlibaba	0.36	$0.30 / $1.50
60	Seed-OSS-36B-InstructByteDance · best of 3 settings	0.36	–
88	Claude 3.7 Sonnet (reasoning on)Anthropic · best of 2 settings	0.35	–
88	Llama 2 70B ChatMeta	0.35	–
88	Mistral Small 3.1 24BMistral AI	0.35	$0.35 / $0.56
88	ChatGPT-4o (2025-03-26)OpenAI	0.35	–
88	Grok 4.20 Beta (0309, non-reasoning)xAI	0.35	–
88	Command R (08-2024)Cohere	0.35	$0.15 / $0.60
88	Gemma 4 26B A4B (reasoning by prefill)Google · best of 2 settings	0.35	$0.10 / $0.30
88	Mistral Medium 3Mistral AI	0.35	$0.40 / $2
88	Qwen3.6-Plus (no reasoning)Alibaba · best of 2 settings	0.35	$0.33 / $1.95
88	Gemini 3.5 Flash (medium reasoning)Google · best of 4 settings	0.35	$1.50 / $9
88	Granite 4.0 H SmallIBM	0.35	–
88	GPT-5 Chat (latest)OpenAI	0.35	–
88	Qwen3-VL-32B-InstructAlibaba	0.35	$0.10 / $0.42
88	Llama 3.3 8B InstructAllura Forge	0.35	–
88	Mixtral 8x22B InstructMistral AI	0.35	$2 / $6
88	Mistral Small 3Mistral AI	0.35	$0.05 / $0.08
88	DeepSeek-V3-0324DeepSeek	0.35	$0.25 / $1
88	Mistral Small 3.2 24BMistral AI	0.35	$0.094 / $0.25
88	Mistral Small (2409)Mistral AI	0.35	–
88	GPT-5.2 ChatOpenAI	0.35	–
88	Qwen3.6-27B (no reasoning)Alibaba · best of 2 settings	0.35	$0.30 / $3.20
88	Nova 2 Lite (no reasoning)Amazon · best of 2 settings	0.35	$0.30 / $2.50
88	Ministral 3 14B ReasoningMistral AI · best of 2 settings	0.35	–
88	Mistral Medium 3.5 (no reasoning)Mistral AI · best of 2 settings	0.35	$1.50 / $7.50
88	Kimi K2.6 (reasoning on)Moonshot AI · best of 2 settings	0.35	$0.95 / $4
88	Claude 3 OpusAnthropic	0.35	–
88	Command ACohere	0.35	$2.50 / $10
88	DeepSeek-V3.2-Exp (no reasoning)DeepSeek · best of 2 settings	0.35	$0.27 / $0.41
88	Devstral Small 2Mistral AI	0.35	–
88	Qwen3-4B-Instruct-2507Alibaba	0.35	–
88	Gemini 2.5 Flash Preview (09-2025, reasoning on)Google · best of 2 settings	0.35	–
88	Gemma 4 12BGoogle · best of 2 settings	0.35	–
88	Ministral 3 8B Reasoning (reasoning by prefill)Mistral AI · best of 3 settings	0.35	–
88	Hy3 Preview (no reasoning)Tencent · best of 2 settings	0.35	$0.18 / $0.60
122	Qwen3-VL-235B-A22B-ThinkingAlibaba	0.34	$0.40 / $4
122	Gemini 3.1 Pro Preview (low reasoning)Google · best of 3 settings	0.34	$2 / $12
122	Llama 3.1 8B InstructMeta	0.34	$0.05 / $0.08
122	Phi-4Microsoft	0.34	$0.07 / $0.14
122	Inflection 3 PiInflection AI	0.34	–
122	Magistral Small 1.2Mistral AI · best of 2 settings	0.34	–
122	Kimi K2.5 (no reasoning)Moonshot AI · best of 2 settings	0.34	$0.57 / $2.85
122	Grok 4.20 (0309, non-reasoning)xAI	0.34	–
122	Command R+ (08-2024)Cohere	0.34	$2.50 / $10
122	Claude 3 HaikuAnthropic	0.34	–
122	Gemini 3.8 Flash (high reasoning)Google · best of 3 settings	0.34	$1.50 / $7.50
122	Kimi-VL-A3B-InstructMoonshot AI	0.34	–
122	EXAONE 4.0 32BLG AI Research	0.34	–
122	Mistral NemoMistral AI	0.34	$0.023 / $0.03
122	GPT-5.1 ChatOpenAI	0.34	–
122	Grok 4.3xAI	0.34	$1.25 / $2.50
122	Qwen3-VL-32B-ThinkingAlibaba	0.34	–
139	Kimi-VL-A3B-Thinking (2506)Moonshot AI	0.33	–
139	Qwen3-30B-A3B-Thinking-2507Alibaba	0.33	$0.20 / $2.40
139	Qwen3-VL-4B-ThinkingAlibaba	0.33	–
139	Qwen3-VL-8B-InstructAlibaba	0.33	$0.12 / $0.46
139	MedGemma 27B TextGoogle	0.33	–
139	Llama 3.2 1B InstructMeta	0.33	$0.027 / $0.20
139	Ministral 3 8B (reasoning by prefill)Mistral AI · best of 2 settings	0.33	$0.15 / $0.15
139	o4-mini (low reasoning)OpenAI · best of 3 settings	0.33	$1.10 / $4.40
139	Qwen3.5-27B (no reasoning)Alibaba · best of 2 settings	0.33	$0.27 / $2.16
139	Qwen3-VL-2B-ThinkingAlibaba	0.33	–
139	Grok 4.1 Fast (reasoning)xAI	0.33	–
139	Grok 4.20 Beta (0309, reasoning)xAI	0.33	–
139	Qwen3.5-2B (reasoning by prefill)Alibaba · best of 2 settings	0.33	–
139	Qwen3.5-35B-A3B (no reasoning)Alibaba · best of 2 settings	0.33	$0.16 / $1.30
139	GPT-4o (2024-05-13)OpenAI	0.33	$5 / $15
139	Ling-1TAnt Group	0.33	–
139	Qwen3-MaxAlibaba	0.33	$0.78 / $3.90
139	Qwen3-VL-8B-ThinkingAlibaba	0.33	$0.18 / $2.10
139	Grok 4 Fast (reasoning)xAI	0.33	–
139	GLM-4.5-AirZ.ai · best of 2 settings	0.33	$0.14 / $0.86
139	Gemini 3.6 Flash (medium reasoning)Google · best of 3 settings	0.33	$1.50 / $7.50
139	Gemini 3.7 Flash (medium reasoning)Google · best of 3 settings	0.33	$1.50 / $7.50
139	GPT-4.1OpenAI	0.33	$2 / $8
139	Falcon-H1 3B InstructTII	0.33	–
139	Grok 4 Fast (non-reasoning)xAI	0.33	–
139	Qwen3.5-397B-A17B (no reasoning)Alibaba	0.33	$0.55 / $3.50
139	Falcon-H1 7B InstructTII	0.33	–
139	Grok 4.20 (0309, reasoning)xAI	0.33	–
167	Ministral 3 14BMistral AI · best of 2 settings	0.32	$0.20 / $0.20
167	Gemma 4 E2B (reasoning by prefill)Google · best of 2 settings	0.32	–
167	Qwen2.5-VL-72B-InstructAlibaba	0.32	$0.80 / $1
167	Falcon-H1 1.5B InstructTII	0.32	–
167	Solar Pro 3Upstage	0.32	$0.15 / $0.60
167	GLM-4-32B-0414Z.ai	0.32	–
167	Qwen3.5-9B (no reasoning)Alibaba · best of 2 settings	0.32	$0.10 / $0.15
167	Qwen3.6-35B-A3B (no reasoning)Alibaba · best of 2 settings	0.32	$0.10 / $1
167	Qwen3-32B (no reasoning)Alibaba · best of 2 settings	0.32	$0.14 / $0.40
167	Qwen3.5-4B (no reasoning)Alibaba · best of 2 settings	0.32	–
167	Kimi Linear 48B A3B InstructMoonshot AI	0.32	–
167	Qwen2.5-7B-InstructAlibaba	0.32	$0.10 / $0.20
167	Qwen2.5-VL-32B-InstructAlibaba	0.32	–
167	Ring-1TAnt Group	0.32	–
167	Grok 4.1 Fast (non-reasoning)xAI	0.32	–
167	Grok 4.5xAI	0.32	$2 / $6
167	Qwen2.5-Coder-7B-InstructAlibaba	0.32	–
167	Qwen3.5-0.8B (no reasoning)Alibaba	0.32	–
167	Olmo 3 32B ThinkAi2	0.32	–
167	DeepSeek-R1-0528DeepSeek	0.32	$0.50 / $2.18
187	Qwen3-30B-A3B-Instruct-2507Alibaba	0.31	$0.09 / $0.30
187	gpt-oss-20b (medium reasoning)OpenAI · best of 3 settings	0.31	$0.03 / $0.15
187	Qwen3-VL-30B-A3B-ThinkingAlibaba	0.31	$0.29 / $1
187	Falcon-H1 1.5B Deep InstructTII	0.31	–
187	Apriel Nemotron 15B ThinkerServiceNow	0.31	–
187	Mistral Small 4 (high reasoning)Mistral AI · best of 2 settings	0.31	$0.15 / $0.60
187	Reka Flash 3Reka AI	0.31	$0.10 / $0.20
187	Qwen3-235B-A22B-Thinking-2507Alibaba	0.31	$0.30 / $3
187	Kimi K2 (0905)Moonshot AI	0.31	$0.60 / $2.50
187	Qwen3-Coder-30B-A3B-InstructAlibaba	0.31	$0.07 / $0.28
187	Qwen3-Next-80B-A3B-InstructAlibaba	0.31	$0.10 / $1.10
187	o3 (high reasoning)OpenAI · best of 3 settings	0.31	$2 / $8
199	Qwen2.5-32B-InstructAlibaba	0.30	–
199	Gemma 2 27BGoogle	0.30	$0.65 / $0.65
199	Gemma 4 E4B (reasoning by prefill)Google · best of 2 settings	0.30	–
199	QwQ-32BAlibaba	0.30	–
199	Qwen2.5-VL-3B-InstructAlibaba	0.30	–
199	DeepSeek-R1DeepSeek	0.30	$0.70 / $2.50
199	InternLM3 8B InstructShanghai AI Lab	0.30	–
199	Qwen2.5-1.5B-InstructAlibaba	0.30	–
199	Inflection 3 ProductivityInflection AI	0.30	–
199	Gemma 2 2BGoogle	0.30	–
199	Gemma 2 9BGoogle	0.30	–
199	Olmo 3 7B InstructAi2	0.30	–
211	Qwen3-Omni-30B-A3B-ThinkingAlibaba	0.29	–
211	Gemma 3 12BGoogle	0.29	$0.05 / $0.15
211	Gemma 3 27BGoogle	0.29	$0.12 / $0.20
211	Gemma 3 4BGoogle	0.29	$0.05 / $0.10
211	Olmo 3 7B ThinkAi2	0.29	–
211	Qwen3-VL-4B-InstructAlibaba	0.29	–
211	Qwen2.5-14B-InstructAlibaba	0.29	–
211	Kimi K2 (0711)Moonshot AI	0.29	$0.57 / $2.30
211	gpt-oss-120b (medium reasoning)OpenAI	0.29	$0.15 / $0.60
211	Qwen3-Next-80B-A3B-ThinkingAlibaba	0.29	$0.15 / $1.20
221	Nemotron 3 Nano 30B A3BNVIDIA · best of 2 settings	0.28	$0.05 / $0.20
221	Nanbeige4-3B-Thinking (2510)Nanbeige	0.28	–
221	Solar 10.7B Instruct v1.0Upstage	0.28	–
221	Qwen3-4B-Thinking-2507Alibaba	0.28	–
221	Qwen3-VL-2B-InstructAlibaba	0.28	–
221	LFM2-8B-A1BLiquid AI	0.28	–
227	LFM2-24B-A2BLiquid AI	0.27	–
227	Nemotron Nano 12B v2 VL (BF16, no reasoning)NVIDIA · best of 2 settings	0.27	–
229	Nanbeige4-3B-Thinking (2511)Nanbeige	0.26	–
230	Qwen3-0.6B (reasoning on)Alibaba	0.25	–

Swipe the table sideways for more columns.

Ranks follow the score as shown, so equal numbers share a rank. Each model is shown at its best setting; show every setting. Results as published by UGI Leaderboard; we do not re-run them.

What it measures

How closely the model matches the writing style of a given example.

What it does not measure

Not brand voice on your own examples: UGI's prompts are private and lean towards creative writing.

Source

UGI Leaderboard · Methodology

947 results from UGI Leaderboard not ranked here · show why

We rank a result only when we can tie it to a specific model you can use. These are left out:

Community fine-tunes and merges: 934
No score published: 13

UGI Leaderboard by DontPlanToEnd, Apache License 2.0.