GPQA (Open LLM Leaderboard)

Reasoning

Source ↗ Higher is better

Rank	Model	Provider	Score
1	Steelskull/L3.3-MS-Nevoria-70b	Steelskull	29.4%
2	Steelskull/L3.3-Nevoria-R1-70b	Steelskull	29.2%
3	Tarek07/Progenitor-V1.1-LLaMa-70B	Tarek07	27.7%
4	rootxhacker/Apollo-70B	rootxhacker	27.6%
5	Sao10K/70B-L3.3-Cirrus-x1	Sao10k	26.6%
6	Triangle104/Set-70b	Triangle104	26.2%
7	Daemontatox/Llama3.3-70B-CogniLink	Daemontatox	26.1%
8	Tarek07/Thalassic-Alpha-LLaMa-70B	Tarek07	25.8%
9	mistralai/Mistral-Large-Instruct-2411	Mistral AI	24.9%
10	OpenBuddy/openbuddy-llama3.3-70b-v24.1-131k	OpenBuddy	24.6%
11	pankajmathur/orca_mini_v8_1_70b	pankajmathur	24.4%
12	JungZoona/T3Q-Qwen2.5-14B-Instruct-1M-e3	JungZoona	22.3%
13	JungZoona/T3Q-qwen2.5-14b-v1.0-e3	JungZoona	22.3%
14	newsbang/Homer-v1.0-Qwen2.5-72B	newsbang	22.1%
15	Sakalti/ultiima-72B	Sakalti	21.9%
16	Sakalti/ultiima-72B-v1.5	Sakalti	21.8%
17	shuttleai/shuttle-3	shuttleai	21.6%
18	Qwen/Qwen2.5-32B	Alibaba	21.6%
19	hotmailuser/Mistral-modelstock-24B	hotmailuser	21.4%
20	Daemontatox/PathFinderAi3.0	Daemontatox	21.1%
21	EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2	EVA-UNIT-01	21.1%
22	CultriX/Qwen2.5-14B-ReasoningMerge	CultriX	21.0%
23	MaziyarPanahi/calme-2.2-rys-78b	MaziyarPanahi	20.9%
24	zetasepic/Qwen2.5-72B-Instruct-abliterated	zetasepic	20.9%
25	NyxKrage/Microsoft_Phi-4	NyxKrage	20.8%
26	rombodawg/Rombos-LLM-V2.6-Nemotron-70b	rombodawg	20.8%
27	Microsoft: Phi 4	Microsoft	20.8%
28	NikolaSigmoid/phi-4-300steps	NikolaSigmoid	20.7%
29	Qwen/Qwen2.5-72B	Alibaba	20.7%
30	MaziyarPanahi/calme-2.3-rys-78b	MaziyarPanahi	20.6%
31	Orion-zhen/phi-4-abliterated	Orion-zhen	20.6%
32	prithivMLmods/Sombrero-Opus-14B-Sm1	prithivMLmods	20.5%
33	NikolaSigmoid/phi-4-14b	NikolaSigmoid	20.5%
34	MaziyarPanahi/calme-2.4-rys-78b	MaziyarPanahi	20.4%
35	MaziyarPanahi/calme-3.2-instruct-78b	MaziyarPanahi	20.4%
36	NikolaSigmoid/phi-4-1steps	NikolaSigmoid	20.2%
37	prithivMLmods/Viper-Coder-v1.1	prithivMLmods	20.1%
38	prithivMLmods/Viper-Coder-v1.6-r999	prithivMLmods	20.1%
39	sometimesanotion/Lamarck-14B-v0.7-Fusion	sometimesanotion	20.1%
40	fblgit/TheBeagle-v2beta-32B-MGS	fblgit	20.1%
41	prithivMLmods/Coma-II-14B	prithivMLmods	20.0%
42	dfurman/CalmeRys-78B-Orpo-v0.1	dfurman	20.0%
43	Baptiste-HUVELLE-10/LeTriomphant2.2_ECE_iLAB	Baptiste-HUVELLE-10	19.9%
44	prithivMLmods/Galactic-Qwen-14B-Exp2	prithivMLmods	19.9%
45	prithivMLmods/Phi-4-Math-IO	prithivMLmods	19.8%
46	rombodawg/Rombos-LLM-V2.5-Qwen-72b	rombodawg	19.8%
47	meta-llama/Meta-Llama-3-70B	Meta	19.7%
48	1024m/PHI-4-Hindi	1024m	19.7%
49	prithivMLmods/Viper-Coder-v1.7-Vsm6	prithivMLmods	19.6%
50	rombodawg/Rombos-LLM-V2.5-Qwen-32b	rombodawg	19.6%

Access this data

JSON GET /api/v1

[
  {
    "rank": 1,
    "model": "Steelskull/L3.3-MS-Nevoria-70b",
    "provider": "Steelskull",
    "score": "29.418344519015662"
  },
  {
    "rank": 2,
    "model": "Steelskull/L3.3-Nevoria-R1-70b",
    "provider": "Steelskull",
    "score": "29.194630872483216"
  },
  {
    "rank": 3,
    "model": "Tarek07/Progenitor-V1.1-LLaMa-70B",
    "provider": "Tarek07",
    "score": "27.740492170022367"
  }
]

Curl

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

CLI brew install rightrank

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

Claude Code Skill Skill

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