Atlas

Models

← All models

Phi-4-reasoning-plus

Microsoft Research · 2025-04-30 · 14.7B parameters

Phi-4-reasoning-plus is Microsoft's open-weight 14B reasoning model, fine-tuned from Phi-4 with supervised chain-of-thought data and additional reinforcement learning. It targets math, science, and coding tasks and emits a reasoning block followed by a summarized answer.

Benchmark scores

BenchmarkScore
FinanceQA - Assumption-Based43.2
FormationEval87.7
Loading Atlas data…