Codesota · Models · microsoft/phi-4microsoft16 results · 3 benchmarks
Model card

microsoft/phi-4.

microsoftopen-weights14.7B params
§ 02 · Benchmarks

Every benchmark microsoft/phi-4 has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalklej-ner-mc61.3%#10/490—source ↗
02Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalbelebele91.6%#17/490—source ↗
03Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalcbd37.3%#29/490—source ↗
04Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalpolemo2-in86.0%#34/490—source ↗
05Polish EQ-BenchNatural Language Processing · Polish Emotional Intelligenceeq-score59.1%#34/101—source ↗
06CPTU-BenchNatural Language Processing · Polish Text Understandingaverage3.3%#47/93—source ↗
07CPTU-BenchNatural Language Processing · Polish Text Understandingsentiment3.7%#48/93—source ↗
08CPTU-BenchNatural Language Processing · Polish Text Understandingtricky-questions2.7%#48/93—source ↗
09CPTU-BenchNatural Language Processing · Polish Text Understandinglanguage-understanding3.5%#48/93—source ↗
10CPTU-BenchNatural Language Processing · Polish Text Understandingphraseology3.2%#52/93—source ↗
11Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalaverage62.6%#55/491—source ↗
12Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generaldyk66.1%#66/489—source ↗
13Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalklej-ner-generative54.1%#83/489—source ↗
14Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalpoquad-open-book67.0%#88/337—source ↗
15Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalppc76.3%#91/490—source ↗
16Open PL LLM LeaderboardNatural Language Processing · Polish LLM Generalpolqa-open-book84.5%#217/489—source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Strengths by area

Where microsoft/phi-4 actually performs.

Natural Language Processing
3
benchmarks
avg rank #60.4
§ 05 · Related models

Other microsoft models scored on Codesota.

Phi-3-medium-4k-instruct
0 results
Phi-3-mini-4k-instruct
0 results
Phi-3-small-8k-instruct
0 results
Phi-3.5-MoE-instruct
0 results
microsoft/Phi-3.5-mini-instruct
3.82B params · 0 results
microsoft/Phi-4-mini-instruct
3.84B params · 0 results
🚧WizardLM-2-7B
0 results
§ 06 · Sources & freshness

Where these numbers come from.

speakleash/open_pl_llm_leaderboard
10
results
SpeakLeash/CPTU-Bench
5
results
SpeakLeash/Polish-EQ-Bench
1
result
16 of 16 rows marked verified.