AnalysisAI ModelsJuly 23, 2026
FinMMEval 2026 multilingual financial QA benchmark released
The benchmark includes Task 1 (multiple-choice) in English, Chinese, Arabic, and Hindi, and Task 2 (short-answer) in English, Chinese, Japanese, Spanish, and Greek. It evaluates models on domain-specific financial knowledge using multilingual evidence.
2 sources
More stories today
- Hugging Face Uses Chinese Model in Self-Defense
- How AI guardrails are impeding the work of offensive cybersecurity researchers
- Fugu-Ultra v1.1 beats Fable 5 by 7.9 points in coding and reasoning
- audio.cpp 0.4 adds Higgs Audio v3 TTS 4B and Fish Audio S2 Pro
- Meta's AI optimism ad uses David Bowie song about apocalypse