Agentic · tools

DeepPlanning leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

As of 2026-10-07, the highest measured score on DeepPlanning is 62.3% by Qwen3.7 Plus. 50 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Qwen3.7 Max	100.0%	estimated ± 9.1 pp, low confidence
2	Qwen3.7 Plus	62.3%	measured
3	Muse Spark 1.1	53.8%	estimated ± 13.1 pp, low confidence
4	Claude Opus 5	52.7%	estimated ± 13.1 pp, low confidence
5	Step 5 Preview	52.6%	estimated ± 13.1 pp, low confidence
6	Kimi K3	52.0%	estimated ± 13.1 pp, low confidence
7	Hy4 preview	51.8%	estimated ± 13.1 pp, low confidence
8	Gemini 3.5 Flash	51.7%	estimated ± 13.1 pp, low confidence
9	Claude Opus 4.8	51.1%	estimated ± 13.1 pp, low confidence
10	Ornith-1.5-397B	50.1%	estimated ± 13.1 pp, low confidence
11	Inkling-Small	49.9%	estimated ± 13.1 pp, low confidence
12	Beam	49.4%	estimated ± 13.1 pp, low confidence
13	Claude Opus 4.7 (Adaptive)	48.8%	estimated ± 13.1 pp, low confidence
14	GLM-5.2	48.5%	estimated ± 13.1 pp, low confidence
15	Kimi K2.7 Code	48.2%	estimated ± 13.1 pp, low confidence
16	Muse Glimmer 30B	47.9%	estimated ± 13.1 pp, low confidence
17	GPT-5.5	47.8%	estimated ± 13.1 pp, low confidence
18	MiniMax M3	47.3%	estimated ± 13.1 pp, low confidence
19	Inkling	47.2%	estimated ± 13.1 pp, low confidence
20	DeepSeek V4 Pro 0813	47.0%	estimated ± 13.1 pp, low confidence
21	GPT-5.4	45.5%	estimated ± 13.1 pp, low confidence
22	Ornith-1.5-35B-A3B	45.3%	estimated ± 13.1 pp, low confidence
23	DeepSeek V4 Flash 0731	44.7%	estimated ± 13.1 pp, low confidence
24	Qwen 3.6 Max (preview)	43.2%	estimated ± 11.8 pp, low confidence
25	Ling 3.0 Flash	42.9%	estimated ± 13.1 pp, low confidence
26	Qwen3.6 Plus	41.5%	measured
27	Solar Pro 4	40.8%	estimated ± 13.1 pp, low confidence
28	Solar Open 2	39.1%	estimated ± 13.1 pp, low confidence
29	GPT-5.4 mini	38.8%	estimated ± 13.1 pp, low confidence
30	GPT-5.4 nano	37.9%	estimated ± 13.1 pp, low confidence
31	Kimi K2.6	37.8%	estimated ± 13.1 pp, low confidence
32	Qwen3.5 397B	37.6%	measured
33	Ornith-1.5-9B	36.9%	estimated ± 13.1 pp, low confidence
34	Mercury 2.5	35.4%	estimated ± 8.3 pp, low confidence
35	Mistral Medium 3.5 128B	35.4%	estimated ± 8.3 pp, low confidence
36	MiMo-V2.5-Pro	35.4%	estimated ± 8.3 pp, low confidence
37	Nemotron 3 Ultra	35.4%	estimated ± 8.3 pp, low confidence
38	GLM-5.1	35.4%	estimated ± 8.3 pp, low confidence
39	LLaDA2.2-flash	32.3%	estimated ± 13.1 pp, low confidence
40	Agents-A1-4B	26.9%	estimated ± 9.1 pp, low confidence
41	Agents-A1	26.4%	estimated ± 9.1 pp, low confidence
42	Claude Opus 4.5	26.4%	measured
43	Claude Sonnet 4.5	26.2%	estimated ± 9.1 pp, low confidence
44	DeepSeek V3.2	26.2%	estimated ± 9.1 pp, low confidence
45	GLM-4.7	26.2%	estimated ± 9.1 pp, low confidence
46	LongCat-Flash-Lite-Sparse	26.2%	estimated ± 9.1 pp, low confidence
47	Qwen3.6-35B-A3B	25.9%	measured
48	Qwen3.6-27B	25.3%	estimated ± 11.8 pp, low confidence
49	Pokee-Isaac 28B	16.9%	estimated ± 8.3 pp, low confidence
50	GLM-5	14.6%	measured
51	Kimi K2.5	14.4%	measured
52	Granite 4.2 30B	13.0%	estimated ± 8.3 pp, low confidence
53	Atria Dawn Preview	13.0%	estimated ± 8.3 pp, low confidence
54	Granite 4.2 3B	13.0%	estimated ± 8.3 pp, low confidence
55	Granite 4.2 8B	13.0%	estimated ± 8.3 pp, low confidence
56	LFM2.5-2.6B	13.0%	estimated ± 8.3 pp, low confidence
57	Nemotron 3.5 Lightning 30B A3B NVFP4	13.0%	estimated ± 8.3 pp, low confidence

DeepPlanning leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General