Microsoft
Phi-4-reasoning-plus 14B
最先端
19.7Kダウンロード340いいねApr 2025公開日33K トークンコンテキストMITライセンス92 卓越品質
Phi-4-reasoning-plus 14B (14.699999809265137B parameters) requires approximately 13.8 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 16 GB of VRAM.
はじめに
— コピー&ペーストでローカル実行Copy-paste commands to run Phi-4-reasoning-plus 14B on your machine.
Run
ollama run phi4-reasoningQuick specs
Parameters14.7B
Architecturedense
Context33K tokens
Modalitytext
Min RAM5.7 GB
Rec. RAM9 GB (Q4_K_M)
LicenseMIT
FamilyPhi
✓ Chat✓ Reasoning
About this model
- •AIME 2025, 2024, 2023, and 2022:: Math olympiad questions
- •GPQA-Diamond:: Complex, graduate-level science questions
- •OmniMath:: Collection of over 4000 olympiad-level math problems with human annotation
- •LiveCodeBench:: Code generation benchmark gathered from competitive coding contests
- •3SAT (3-literal Satisfiability Problem) and TSP (Traveling Salesman Problem):: Algorithmic problem solving
関連モデル
おすすめ
最適なハードウェア
Phi-4-reasoning-plus 14Bのおすすめ
このモデルを実行
量子化オプション
量子化レベル別VRAM推定値
No hardware detected — fit column shows raw VRAM estimates
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 5.7 GB | Low | — |
Q3_K_S | 3 | 7.2 GB | Low | — |
NVFP4 | 4 | 8.2 GB | Medium | — |
Q4_K_M | 4 | 9.0 GB | Medium | — |
Q5_K_M | 5 | 10.6 GB | High | — |
Q6_K | 6 | 12.1 GB | High | — |
Q8_0 | 8 | 15.7 GB | Very High | — |
F16 | 16 | 30.1 GB | Maximum | — |
Quality benchmarks
Phi-4-reasoning-plus 14B benchmark scores
Coding
SWE-bench Verified—
HumanEval+92.3%
Aider Polyglot—
LiveCodeBench53.1%
Reasoning
MMLU-Pro76.0%
GPQA Diamond68.9%
MATH-50095.4%
ARC Challenge—
General
Chatbot Arena—
IFEval84.9%
Source: official · 2025-04-30
ハードウェア互換性
全ハードウェアの適合度推定
Computing compatibility...
メモリ内訳
Reference: RTX 2060 6GB
Weights9.0 GB
KV Cache3.1 GB
Runtime1.2 GB
Headroom0.6 GB
よくある質問
FAQ — Phi-4-reasoning-plus 14B
関連項目