PhivsMiniMax

Comparing Microsoft and MiniMax. Average Podium Score across top models: Phi (58.4) vs MiniMax (58.3). Flagship showdown: Phi 3 Medium 4k Instruct (60.0) vs Minimax M2.5 (70.0).

Phi (Microsoft)

Highly capable Small Language Models (SLMs) from Microsoft Research.

Avg Score: 58.4
Licensing: Open Weights Available

MiniMax (MiniMax)

High-throughput reasoning and multimodal models from MiniMax.

Top Model: Minimax M2.5
Avg Score: 58.3
Licensing: Proprietary API

Top Models Showdown

RankModelLabScoreCodingReasoningSpeedPrice / 1M
#1Minimax M2.5MiniMax70.050 t/s$3/M
#2Minimax M2.1 PreviewMiniMax69.050 t/s$3/M
#3Minimax M1MiniMax68.050 t/s$3/M
#4Phi 3 Medium 4k InstructMicrosoft60.050 t/s$3/M
#5Wizardlm 70bMicrosoft59.050 t/s$3/M
#6Phi 3 Small 8k InstructMicrosoft59.050 t/s$3/M
#7Wizardlm 13bMicrosoft57.050 t/s$3/M
#8Phi 3 Mini 4k Instruct June 2024Microsoft57.050 t/s$3/M
#9MiniMax-M2.7MiniMax45.055.060.040 t/s$2/M
#10MiniMax M3MiniMax39.638.764.893 t/s$1.2/M
Looking for more ecosystem comparisons? Phi vs Claude · Phi vs ChatGPT / GPT · Phi vs Kimi · Phi vs DeepSeek · Phi vs Gemini · Phi vs Qwen
0 / 4 Models Selected