Multilingual

MAXIFE leaderboard

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

As of 2026-10-10, the highest measured score on MAXIFE is 89.2% by Qwen3.7 Max. 13 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Qwen3.7 Max	89.2%	measured
2	Qwen3.7 Plus	88.8%	measured
3	Qwen3.5 397B	88.2%	measured
4	K-EXAONE 2.0	84.0%	estimated ± 4.7 pp, medium confidence
5	Qwen3.6 Plus	83.5%	estimated ± 2.2 pp, medium confidence
6	Claude Opus 4.5	79.2%	measured
7	Nemotron 3 Ultra	77.3%	estimated ± 6.5 pp, low confidence
8	Qwen3.5-122B-A10B	74.8%	estimated ± 6.5 pp, low confidence
9	Qwen3.5-27B	74.8%	estimated ± 6.5 pp, low confidence
10	Kimi K2.5	72.8%	measured
11	Qwen3 235B 2507 (Reasoning)	71.2%	estimated ± 6.5 pp, low confidence
12	Qwen3.5-35B-A3B	71.2%	estimated ± 6.5 pp, low confidence
13	GLM-5	69.7%	estimated ± 2.2 pp, low confidence
14	Qwen3 235B 2507	66.4%	estimated ± 6.5 pp, low confidence
15	GPT-4.1	47.2%	estimated ± 6.5 pp, low confidence
16	DeepSeek V3 0324	41.3%	estimated ± 6.5 pp, low confidence
17	GPT-4o	17.2%	estimated ± 6.5 pp, low confidence
18	Phi-4	0.0%	estimated ± 6.5 pp, low confidence

MAXIFE leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General