MuSR

Reasoning

Multistep Soft Reasoning over ~1000-word narratives (murder mysteries, object placement)

Higher is better

Rank	Model	Provider	Score
1	JungZoona/T3Q-Qwen2.5-14B-Instruct-1M-e3	JungZoona	38.7%
2	JungZoona/T3Q-qwen2.5-14b-v1.0-e3	JungZoona	38.7%
3	MaziyarPanahi/calme-3.2-instruct-78b	MaziyarPanahi	38.5%
4	MaziyarPanahi/calme-3.1-instruct-78b	MaziyarPanahi	36.5%
5	dfurman/CalmeRys-78B-Orpo-v0.1	dfurman	36.4%
6	MaziyarPanahi/calme-2.4-rys-78b	MaziyarPanahi	34.6%
7	LeroyDyer/_Spydaz_Web_AI_AGI_R1_OmG_Coder	LeroyDyer	32.4%
8	Triangle104/DS-R1-Distill-Q2.5-14B-Harmony_V0.1	Triangle104	31.9%
9	win10/miscii-14b-1M-0128	win10	28.8%
10	deepseek-ai/DeepSeek-R1-Distill-Qwen-14B	DeepSeek	28.7%
11	prithivMLmods/Galactic-Qwen-14B-Exp2	prithivMLmods	28.5%
12	prithivMLmods/Coma-II-14B	prithivMLmods	28.1%
13	prithivMLmods/Gauss-Opus-14B-R999	prithivMLmods	27.8%
14	prithivMLmods/Sombrero-Opus-14B-Sm1	prithivMLmods	27.2%
15	prithivMLmods/Viper-Coder-v1.6-r999	prithivMLmods	26.2%
16	prithivMLmods/Viper-Coder-v1.1	prithivMLmods	26.2%
17	johnsutor/Llama-3-8B-Instruct_dare_ties-density-0.9	johnsutor	25.9%
18	microsoft/Orca-2-13b	Microsoft	25.8%
19	prithivMLmods/Sombrero-Opus-14B-Sm4	prithivMLmods	25.8%
20	CultriX/Qwen2.5-14B-ReasoningMerge	CultriX	25.6%
21	prithivMLmods/Condor-Opus-14B-Exp	prithivMLmods	25.4%
22	pankajmathur/orca_mini_v3_70b	pankajmathur	25.1%
23	mlabonne/Hermes-3-Llama-3.1-70B-lorablated	mlabonne	24.7%
24	rombodawg/Rombos-LLM-V2.5-Qwen-32b	rombodawg	24.7%
25	prithivMLmods/Sombrero-Opus-14B-Sm2	prithivMLmods	24.5%
26	fblgit/TheBeagle-v2beta-32B-MGS	fblgit	24.5%
27	sakaltcommunity/novablast-preview	Sakalti	24.5%
28	allenai/Llama-3.1-Tulu-3-70B-SFT	Ai2	24.5%
29	johnsutor/Llama-3-8B-Instruct_breadcrumbs-density-0.7-gamma-0.01	johnsutor	24.5%
30	hotmailuser/RombosBeagle-v2beta-MGS-32B	hotmailuser	24.5%
31	Triangle104/Phi-4-AbliteratedRP	Triangle104	24.4%
32	FINGU-AI/QwQ-Buddy-32B-Alpha	FINGU-AI	24.4%
33	allenai/Llama-3.1-Tulu-3-70B	Ai2	24.3%
34	allenai/Llama-3.1-Tulu-3-70B	Meta	24.3%
35	Sakalti/Oxyge1-33B	Sakalti	24.3%
36	freewheelin/free-solar-evo-v0.11	freewheelin	24.3%
37	johnsutor/Llama-3-8B-Instruct_breadcrumbs-density-0.9-gamma-0.01	johnsutor	24.2%
38	Danielbrdz/Barcenas-14b-phi-4	Danielbrdz	24.2%
39	pankajmathur/orca_mini_v7_72b	pankajmathur	24.2%
40	Sakalti/ultiima-32B	Sakalti	24.1%
41	microsoft/Orca-2-7b	Microsoft	24.1%
42	Krystalan/DRT-o1-7B	Krystalan	24.1%
43	freewheelin/free-solar-evo-v0.13	freewheelin	24.1%
44	NikolaSigmoid/phi-4-14b	NikolaSigmoid	24.0%
45	deepseek-ai/deepseek-llm-67b-chat	DeepSeek	23.9%
46	jpacifico/Chocolatine-2-14B-Instruct-v2.0	jpacifico	23.9%
47	rootxhacker/Apollo_v2-32B	rootxhacker	23.8%
48	arcee-ai/Arcee-Blitz	Arcee AI	23.8%
49	NyxKrage/Microsoft_Phi-4	NyxKrage	23.8%
50	Microsoft: Phi 4	Microsoft	23.8%

Access this data

JSON GET /api/v1

[
  {
    "rank": 1,
    "model": "JungZoona/T3Q-Qwen2.5-14B-Instruct-1M-e3",
    "provider": "JungZoona",
    "score": "38.688020833333326"
  },
  {
    "rank": 2,
    "model": "JungZoona/T3Q-qwen2.5-14b-v1.0-e3",
    "provider": "JungZoona",
    "score": "38.688020833333326"
  },
  {
    "rank": 3,
    "model": "MaziyarPanahi/calme-3.2-instruct-78b",
    "provider": "MaziyarPanahi",
    "score": "38.52890624999999"
  }
]

Curl

$ curl -s https://rightrank.ai/api/v1/rankings/musr.json | jq '.[:3]'

CLI brew install rightrank

$ rightrank top musr --limit 5
$ rightrank compare claude-opus-4.6 gpt-5.4-pro --category musr
$ rightrank value musr  # best performance per dollar

Claude Code Skill Skill

# In Claude Code, use the RightRank skill:
/rightrank best model for musr
/rightrank compare top 5 by value
/rightrank cheapest model scoring above 80 on musr