Atlas

Models

← All models

Qwen-Plus (2025-01-25)

Alibaba

Qwen-Plus 2025-01-25 is a dated, text-only snapshot of Alibaba Cloud's proprietary Qwen-Plus API model from the Qwen2.5 era. Alibaba describes Qwen-Plus as balancing capability, speed, and cost for moderately complex tasks; this snapshot has no thinking mode, and its official profile reports a 128K maximum input and 131.07K context only at rounded precision.

Benchmark scores

BenchmarkScore
GPQA Diamond48.1
MATH Level 565.3
OTIS Mock AIME 2024-202517.8
Phare Average Safety68.8
Phare Bias Resistance55.7
Phare Debunking96.3
Phare Debunking (English)96.7
Phare Debunking (French)95.6
Phare Debunking (Spanish)96.7
Phare Encoding Jailbreaks21.9
Phare Encoding Jailbreaks (English)23.1
Phare Encoding Jailbreaks (French)24.7
Phare Encoding Jailbreaks (Spanish)17.7
Phare Factuality57.8
Phare Factuality (English)73.3
Phare Factuality (French)51.4
Phare Factuality (Spanish)48.6
Phare Framing Jailbreaks59.2
Phare Framing Jailbreaks (English)64.5
Phare Framing Jailbreaks (French)57.5
Phare Framing Jailbreaks (Spanish)55.7
Phare Hallucination Resistance75.6
Phare Harm Resistance94.1
Phare Harmful Misguidance94.1
Phare Harmful Misguidance (English)95.9
Phare Harmful Misguidance (French)92.8
Phare Harmful Misguidance (Spanish)93.7
Phare Jailbreak Resistance50.0
Phare LLM Benchmark68.8
Phare Misinformation77.9
Loading Atlas data…