MMLU-Pro (Open LLM Leaderboard)

General

Source ↗ Higher is better

Rank	Model	Provider	Score
1	MaziyarPanahi/calme-3.2-instruct-78b	MaziyarPanahi	70.0%
2	MaziyarPanahi/calme-3.1-instruct-78b	MaziyarPanahi	68.7%
3	dfurman/CalmeRys-78B-Orpo-v0.1	dfurman	66.8%
4	MaziyarPanahi/calme-2.4-rys-78b	MaziyarPanahi	66.7%
5	arcee-ai/Arcee-Blitz	Arcee AI	57.3%
6	newsbang/Homer-v1.0-Qwen2.5-72B	newsbang	57.2%
7	Sakalti/ultiima-72B-v1.5	Sakalti	56.2%
8	Qwen/Qwen2.5-72B	Alibaba	55.2%
9	Aryanne/QwentileSwap	Aryanne	55.0%
10	rombodawg/Rombos-LLM-V2.5-Qwen-72b	rombodawg	54.8%
11	rombodawg/Rombos-LLM-V2.5-Qwen-32b	rombodawg	54.6%
12	sakaltcommunity/novablast-preview	Sakalti	54.6%
13	fblgit/TheBeagle-v2beta-32B-MGS	fblgit	54.6%
14	Sakalti/ultiima-32B	Sakalti	54.6%
15	Sakalti/Oxyge1-33B	Sakalti	54.5%
16	hotmailuser/RombosBeagle-v2beta-MGS-32B	hotmailuser	54.5%
17	Sakalti/ultiima-72B	Sakalti	54.5%
18	JungZoona/T3Q-qwen2.5-14b-v1.0-e3	JungZoona	54.3%
19	JungZoona/T3Q-Qwen2.5-14B-Instruct-1M-e3	JungZoona	54.3%
20	maldv/Qwentile2.5-32B-Instruct	maldv	54.2%
21	zetasepic/Qwen2.5-72B-Instruct-abliterated	zetasepic	54.1%
22	rootxhacker/Apollo_v2-32B	rootxhacker	54.1%
23	Baptiste-HUVELLE-10/LeTriomphant2.2_ECE_iLAB	Baptiste-HUVELLE-10	53.9%
24	raphgg/test-2.5-72B	raphgg	53.7%
25	EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2	EVA-UNIT-01	53.5%
26	fluently-lm/FluentlyLM-Prinum	fluently-lm	53.4%
27	Qwen/Qwen2.5-32B	Alibaba	53.4%
28	rubenroy/Gilgamesh-72B	rubenroy	53.4%
29	Saxo/Linkbricks-Horizon-AI-Avengers-V1-32B	Saxo	53.3%
30	FINGU-AI/RomboUltima-32B	FINGU-AI	53.2%
31	tomasmcm/sky-t1-coder-32b-flash	tomasmcm	53.1%
32	Saxo/Linkbricks-Horizon-AI-Avengers-V5-32B	Saxo	52.9%
33	Daemontatox/PathFinderAi3.0	Daemontatox	52.9%
34	Saxo/Linkbricks-Horizon-AI-Avengers-V4-32B	Saxo	52.8%
35	Qwen/Qwen2-72B	Alibaba	52.6%
36	maldv/Awqward2.5-32B-Instruct	maldv	52.5%
37	CombinHorizon/huihui-ai-abliterated-Qwen2.5-32B-Inst-BaseMerge-TIES	CombinHorizon	52.5%
38	Saxo/Linkbricks-Horizon-AI-Avengers-V2-32B	Saxo	52.4%
39	Qwen/Qwen2-VL-72B-Instruct	Alibaba	52.4%
40	shuttleai/shuttle-3	shuttleai	52.4%
41	prithivMLmods/Galactic-Qwen-14B-Exp2	prithivMLmods	52.1%
42	CombinHorizon/zetasepic-abliteratedV2-Qwen2.5-32B-Inst-BaseMerge-TIES	CombinHorizon	52.1%
43	Qwen/QwQ-32B-Preview	Alibaba	52.0%
44	Saxo/Linkbricks-Horizon-AI-Avengers-V6-32B	Saxo	51.9%
45	Qwen/Qwen2.5-32B-Instruct	Alibaba	51.9%
46	Saxo/Linkbricks-Horizon-AI-Avengers-V3-32B	Saxo	51.8%
47	tanliboy/lambda-qwen2.5-32b-dpo-test	tanliboy	51.7%
48	Qwen/Qwen2.5-72B-Instruct	Alibaba	51.4%
49	pankajmathur/orca_mini_v7_72b	pankajmathur	51.4%
50	zetasepic/Qwen2.5-32B-Instruct-abliterated-v2	zetasepic	51.4%

Access this data

JSON GET /api/v1

[
  {
    "rank": 1,
    "model": "MaziyarPanahi/calme-3.2-instruct-78b",
    "provider": "MaziyarPanahi",
    "score": "70.03361406619385"
  },
  {
    "rank": 2,
    "model": "MaziyarPanahi/calme-3.1-instruct-78b",
    "provider": "MaziyarPanahi",
    "score": "68.72229609929079"
  },
  {
    "rank": 3,
    "model": "dfurman/CalmeRys-78B-Orpo-v0.1",
    "provider": "dfurman",
    "score": "66.80149231678487"
  }
]

Curl

$ curl -s https://rightrank.ai/api/v1/rankings/mmlu-pro-open-llm-leaderboard.json | jq '.[:3]'

CLI brew install rightrank

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

Claude Code Skill Skill

# In Claude Code, use the RightRank skill:
/rightrank best model for mmlu-pro-open-llm-leaderboard
/rightrank compare top 5 by value
/rightrank cheapest model scoring above 80 on mmlu-pro-open-llm-leaderboard