BigO(Bench) - Space Complexity Prediction: leaderboard

Each model, zero-shot, must name the space complexity of a given human Code Contests solution; all@1 counts a problem only when every one of its complexity classes is named correctly; each model ran on the 308 problems (636 solutions) of the test set, 20 samples each. These are the original protocol's numbers (paper v1/v2, March 2025, problem statements with their input bounds); the paper's 2026 v3 drops the bounds and re-measures, and is not this board. Llama 3.1 405B, which generated the input parsers that decide which problems enter the test set, is not ranked. Unlike HumanEval-style correctness boards it scores whether code meets a stated algorithmic complexity.

Metric: all@1 (%). Source: facebookresearch.github.io. Saturation forecast: Around 2030. 16 models tracked. Updated 17 Nov 2026.

Top models

#ModelScoreOverall rank
1Gemma 3 27B (IT)13.2#1339
2Qwen 2.5 Coder 32B Instruct12.6#1597
3openhands-lm-32B-v0.112.4#2237
4Llama 3.1 70B Instruct11.9#1577
5DeepSeek V3 (0324)11.9#722
6QwQ-32B11.6#908
7DeepSeek V311.2#734
8Llama 3.3 70B Instruct10.9#1387
9Codestral-22B-v0.110.6#2680
10DeepSeek R1 Distill Llama 70B10.4#1669
11Llama 3.1 Nemotron Ultra 253B v110.3#1014
12Llama 4 Maverick Instruct FP88.9#1022
13DeepSeek-Coder-V2-Instruct-07248.2#1103
14DeepSeek R1 Distill Qwen 32B8.1#1806
15Llama 3.1 8B Instruct7.9#4576

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=bigo-bench-space-complexity-prediction · How It Works · Data refreshed daily, snapshot 2026-11-17.