BBH

Reasoning

BIG-Bench Hard — 23 of the hardest BIG-Bench tasks

Higher is better

Rank	Model	Provider	Score
1	qingy2024/Benchmaxx-Llama-3.2-1B-Instruct	qingy2024	76.7%
2	JungZoona/T3Q-qwen2.5-14b-v1.0-e3	JungZoona	65.5%
3	JungZoona/T3Q-Qwen2.5-14B-Instruct-1M-e3	JungZoona	65.5%
4	shuttleai/shuttle-3	shuttleai	64.1%
5	IntervitensInc/internlm2_5-20b-llamafied	IntervitensInc	63.5%
6	Sakalti/ultiima-72B-v1.5	Sakalti	63.4%
7	internlm/internlm2_5-20b-chat	Shanghai AI Lab	62.8%
8	MaziyarPanahi/calme-3.2-instruct-78b	MaziyarPanahi	62.6%
9	MaziyarPanahi/calme-3.1-instruct-78b	MaziyarPanahi	62.4%
10	newsbang/Homer-v1.0-Qwen2.5-72B	newsbang	62.3%
11	MaziyarPanahi/calme-2.4-rys-78b	MaziyarPanahi	62.2%
12	raphgg/test-2.5-72B	raphgg	62.2%
13	dfurman/CalmeRys-78B-Orpo-v0.1	dfurman	61.9%
14	Qwen/Qwen2.5-72B-Instruct	Alibaba	61.9%
15	rubenroy/Gilgamesh-72B	rubenroy	61.8%
16	MaziyarPanahi/calme-2.2-qwen2.5-72b	MaziyarPanahi	61.8%
17	MaziyarPanahi/calme-2.1-qwen2.5-72b	MaziyarPanahi	61.7%
18	Baptiste-HUVELLE-10/LeTriomphant2.2_ECE_iLAB	Baptiste-HUVELLE-10	61.6%
19	rombodawg/Rombos-LLM-V2.5-Qwen-72b	rombodawg	61.3%
20	Sakalti/ultiima-72B	Sakalti	61.1%
21	huihui-ai/Qwen2.5-72B-Instruct-abliterated	huihui-ai	60.5%
22	prithivMLmods/Galactic-Qwen-14B-Exp2	prithivMLmods	59.9%
23	zetasepic/Qwen2.5-72B-Instruct-abliterated	zetasepic	59.9%
24	mlabonne/BigQwen2.5-52B-Instruct	mlabonne	59.8%
25	MaziyarPanahi/calme-2.3-rys-78b	MaziyarPanahi	59.6%
26	fluently-lm/FluentlyLM-Prinum	fluently-lm	59.5%
27	MaziyarPanahi/calme-2.1-rys-78b	MaziyarPanahi	59.5%
28	MaziyarPanahi/calme-2.2-rys-78b	MaziyarPanahi	59.3%
29	EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2	EVA-UNIT-01	59.1%
30	dnhkng/RYS-XLarge	dnhkng	58.8%
31	dnhkng/RYS-XLarge-base	dnhkng	58.7%
32	ssmits/Qwen2.5-95B-Instruct	ssmits	58.5%
33	rootxhacker/Apollo_v2-32B	rootxhacker	58.3%
34	rombodawg/Rombos-LLM-V2.5-Qwen-32b	rombodawg	58.3%
35	sakaltcommunity/novablast-preview	Sakalti	58.2%
36	hotmailuser/RombosBeagle-v2beta-MGS-32B	hotmailuser	58.1%
37	Sakalti/ultiima-32B	Sakalti	58.1%
38	Saxo/Linkbricks-Horizon-AI-Avengers-V2-32B	Saxo	58.1%
39	fblgit/TheBeagle-v2beta-32B-MGS	fblgit	58.1%
40	Sakalti/Oxyge1-33B	Sakalti	58.0%
41	anthracite-org/magnum-v2-72b	Anthracite Org	57.9%
42	Aryanne/QwentileSwap	Aryanne	57.7%
43	alpindale/magnum-72b-v1	alpindale	57.7%
44	anthracite-org/magnum-v1-72b	Anthracite Org	57.7%
45	Saxo/Linkbricks-Horizon-AI-Avengers-V1-32B	Saxo	57.6%
46	Undi95/MG-FinalMix-72B	Undi95	57.5%
47	Qwen/Qwen2-72B-Instruct	Alibaba	57.5%
48	dfurman/Qwen2-72B-Orpo-v0.1	dfurman	57.4%
49	MaziyarPanahi/calme-2.1-qwen2-72b	MaziyarPanahi	57.3%
50	VAGOsolutions/Llama-3.1-SauerkrautLM-70b-Instruct	VAGOsolutions	57.2%

Access this data

JSON GET /api/v1

[
  {
    "rank": 1,
    "model": "qingy2024/Benchmaxx-Llama-3.2-1B-Instruct",
    "provider": "qingy2024",
    "score": "76.69996850001888"
  },
  {
    "rank": 2,
    "model": "JungZoona/T3Q-qwen2.5-14b-v1.0-e3",
    "provider": "JungZoona",
    "score": "65.46659667305961"
  },
  {
    "rank": 3,
    "model": "JungZoona/T3Q-Qwen2.5-14B-Instruct-1M-e3",
    "provider": "JungZoona",
    "score": "65.46659667305961"
  }
]

Curl

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

CLI brew install rightrank

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

Claude Code Skill Skill

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