benchgap
Agentic · tools

DeepPlanning leaderboard

As of 2026-11-21, the highest measured score on DeepPlanning is 62.3% by Qwen3.7 Plus. 59 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Qwen3.7 Max100.0%estimated ± 9.1 pp, low confidence
2Qwen3.7 Plus62.3%measured
3Sage-A1.558.5%estimated ± 9.1 pp, low confidence
4Muse Spark Open54.3%estimated ± 13.1 pp, low confidence
5Muse Spark 1.153.8%estimated ± 13.1 pp, low confidence
6Claude Opus 552.7%estimated ± 13.1 pp, low confidence
7Step 5 Preview52.6%estimated ± 13.1 pp, low confidence
8Inkling 2 Small52.4%estimated ± 13.1 pp, low confidence
9ERNIE 5.5 Flash Preview52.2%estimated ± 13.1 pp, low confidence
10Kimi K352.0%estimated ± 13.1 pp, low confidence
11Hy4 preview51.8%estimated ± 13.1 pp, low confidence
12Gemini 3.5 Flash51.7%estimated ± 13.1 pp, low confidence
13LongCat-2.551.5%estimated ± 13.1 pp, low confidence
14Ling 3.1 Flash51.5%estimated ± 13.1 pp, low confidence
15Inkling 2 Mini51.3%estimated ± 13.1 pp, low confidence
16Claude Opus 4.851.1%estimated ± 13.1 pp, low confidence
17Beam 1.0 501B-A23B FP850.8%estimated ± 13.1 pp, low confidence
18Beam 1.0 501B-A23B50.4%estimated ± 13.1 pp, low confidence
19Beam 1.0 501B-A23B NVFP450.3%estimated ± 13.1 pp, low confidence
20Ornith-1.5-397B50.1%estimated ± 13.1 pp, low confidence
21Inkling-Small49.9%estimated ± 13.1 pp, low confidence
22Claude Opus 4.748.8%estimated ± 13.1 pp, low confidence
23GLM-5.248.5%estimated ± 13.1 pp, low confidence
24Kimi K2.7 Code48.2%estimated ± 13.1 pp, low confidence
25Ling 3.0 Flash Sante48.0%estimated ± 13.1 pp, low confidence
26Muse Glimmer 30B47.9%estimated ± 13.1 pp, low confidence
27GPT-5.547.8%estimated ± 13.1 pp, low confidence
28MiniMax M347.3%estimated ± 13.1 pp, low confidence
29Inkling47.2%estimated ± 13.1 pp, low confidence
30DeepSeek V4 Pro47.0%estimated ± 13.1 pp, low confidence
31Ripple-65B Preview46.0%estimated ± 13.1 pp, low confidence
32GPT-5.445.5%estimated ± 13.1 pp, low confidence
33Ornith-1.5-35B-A3B45.3%estimated ± 13.1 pp, low confidence
34DeepSeek V4 Flash 073144.7%estimated ± 13.1 pp, low confidence
35Qwen 3.6 Max (preview)43.2%estimated ± 11.8 pp, low confidence
36Ling 3.0 Flash42.9%estimated ± 13.1 pp, low confidence
37Qwen3.6 Plus41.5%measured
38Spark-X2.6-4B41.3%estimated ± 13.1 pp, low confidence
39Solar Pro 440.8%estimated ± 13.1 pp, low confidence
40Solar Open 239.1%estimated ± 13.1 pp, low confidence
41GPT-5.4 mini38.8%estimated ± 13.1 pp, low confidence
42GPT-5.4 nano37.9%estimated ± 13.1 pp, low confidence
43Kimi K2.637.8%estimated ± 13.1 pp, low confidence
44Nemotron 3.5 Bolt 74B-A9B BF1637.7%estimated ± 13.1 pp, low confidence
45Qwen3.5 397B37.6%measured
46Ornith-1.5-9B36.9%estimated ± 13.1 pp, low confidence
47Nemotron 3.5 Bolt 74B-A9B35.8%estimated ± 13.1 pp, low confidence
48MiMo-V2.5-Pro35.4%estimated ± 8.3 pp, low confidence
49Nemotron 3 Ultra35.4%estimated ± 8.3 pp, low confidence
50GLM-5.135.4%estimated ± 8.3 pp, low confidence
51LFM3-1.7B32.9%estimated ± 8.3 pp, low confidence
52LLaDA2.2-flash32.3%estimated ± 13.1 pp, low confidence
53MiniMax M2.730.4%estimated ± 8.3 pp, low confidence
54Agents-A1-4B26.9%estimated ± 9.1 pp, low confidence
55Agents-A126.4%estimated ± 9.1 pp, low confidence
56Claude Opus 4.526.4%measured
57LongCat-Flash-Lite-Sparse26.2%estimated ± 9.1 pp, low confidence
58DeepSeek V3.226.2%estimated ± 9.1 pp, low confidence
59Claude Sonnet 4.526.2%estimated ± 9.1 pp, low confidence
60GLM-4.726.2%estimated ± 9.1 pp, low confidence
61Qwen3.6-35B-A3B25.9%measured
62Qwen3.6-27B25.3%estimated ± 11.8 pp, low confidence
63Pokee-Isaac 28B16.9%estimated ± 8.3 pp, low confidence
64GLM-514.6%measured
65Kimi K2.514.4%measured
66Granite 4.2 30B13.0%estimated ± 8.3 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General