de Microsoft 14B de pesos abiertos modelo trained extensively en alto-calidad synthetic datasets un deliver matemáticas y codificación density rivaling grande modelos.
Technical Specifications
- Provider
- Microsoft
- Category
- open-source
- Release Date
- 2024-12
- Batch API Discount
- Standard rate
- Consumer Subscription
- Free / Self-Hosted
- Reasoning Effort Control
- Fixed
- Structured Outputs
- Yes
- Output Modalities
- text
- Official Documentation
- Visit Official Docs ↗
Data Privacy, Licensing & Deployment Governance
Official Provider Release Intelligence, API Quotas & Agentic Specs
- Official API Model ID (SDK String)
- phi-4
- Supported SDK & API Protocol
- OpenAI-Compatible REST / vLLM / SGLang / HuggingFace TGI
- Lifecycle & Deprecation Status
- Legacy Stable (Upgrade path available in Microsoft lineup)
- Production Rate Limits (RPM & TPM)
- Unlimited (Hardware-bound on self-hosted GPU) / 10,000 RPM Cloud · Unlimited local throughput / 10M+ TPM Managed Cloud
- Long-Context Pricing Tier (>128K / >200K)
- Free across full context window (only KV-cache VRAM scales)
- Prompt Cache Write & TTL Policy
- Automatic prefix KV-caching in vLLM / SGLang (0 cost)
- Computer Use & Browser GUI Agent
- API / CLI Tool-Use & DOM Agent ready (no raw pixel desktop click loop)
- MCP & Parallel Tool Calling
- Native MCP Client/Server ready · Parallel function calling · 100% Strict JSON Schema
- Web Search Grounding & Code Sandbox
- Supported via external Tavily/Exa Search & E2B/Docker Sandbox tools
- Fine-Tuning & Distillation Support
- Full Weights SFT, LoRA, QLoRA, DPO & GRPO RL (Unsloth / Axolotl / Llama-Factory)
- Multimodal Tokenization Billing
- Text BPE (~4 chars/token)
- Weights Precision & Quantization
- BF16 / FP16, FP8 Dynamic, INT4 AWQ / GPTQ, GGUF (Q4_K_M / Q8_0)
- Minimum GPU VRAM Requirement
- 9 GB VRAM (4-bit Quantized) · 30 GB VRAM (Native BF16)
- Supported Inference Engines
- vLLM, SGLang, Ollama, llama.cpp, TensorRT-LLM, HuggingFace TGI
- System Card & Safety Framework
- Open-Weights System Card + Customizable Guardrail Layer (Llama Guard / ShieldGemma)
- Enterprise Compliance (SOC2 / HIPAA / ZDR)
- 100% Air-Gapped On-Premises / Private VPC · Full HIPAA, GDPR & SOC2 Control
Phi-4 Cost, Context, Hardware & API Integration
Open-weights or free-tier model: $0 API license cost when self-hosted.
Same Monthly Workload Cost Across Popular Reference Models
Live Prompt Paste Tokenizer & Cost Counter
Paste real system instructions, JSON schemas, or code to measure BPE token counts across tokenizers.
Hybrid Prompt Routing Cascade Simulator
Route routine queries to a fast lightweight model ($0.15/1M) and escalate only complex reasoning turns to Phi-4:
Benchmark Scores & 6-Axis Capability Radar
- GPQA Diamond
- 55%
- MMLU-Pro
- 78.5%
- HumanEval
- 88%
- Chatbot Arena Elo
- 1305pts
Strengths
- Remarkable STEM y codificación rendimiento packed en un compacto 14B parámetro huella
- Easily deployable en único desktop GPUs (RTX 4070/4080) y Apple Silicon Macs
- de pesos abiertos con permissive MIT licensing
Limitations & Trade-offs
- Relatively compacto 16K ventana de contexto limits grande documento ingesta
- Max salida tokens capped un 4096 tokens
Editorial Assessment
de Microsoft masterclass en pequeño modelo de lenguaje eficiencia, proving que datos calidad triumphs sobre escala.
Best for: local perimetral desarrollador asistentes, offline matemáticas problem-solving, y efficient workstation tareas
Compare Phi-4
Frequently Asked Questions about Phi-4
How much does the Phi-4 API cost per 1M tokens?+
Phi-4 by Microsoft is priced at $0 per 1M input tokens and $0 per 1M output tokens. With prompt caching enabled, cached input tokens cost $0 per 1M tokens, and asynchronous Batch API requests receive a 0% discount.
Should I use a $20/month subscription or the Phi-4 pay-as-you-go API?+
A flat $20/month consumer subscription is predictable for daily interactive chat and manual coding. However, if you are building software, running automated workflows, or sending low-to-moderate volume with prompt caching enabled, the pay-as-you-go API is typically far more cost-effective and offers higher rate limits without manual UI caps.
How many words and pages fit in Phi-4's context window?+
Phi-4 features a 16K tokens context window, which equals approximately 12,288 English words or ~33 standard single-spaced pages in a single prompt. Its maximum output generation per turn is 4,096 tokens.
Can I run Phi-4 locally on my own GPU (Open Weights)?+
Yes, Phi-4 (14B) releases open weights that can be self-hosted locally or on private cloud GPUs using vLLM, Ollama, SGLang, or llama.cpp.
Does Microsoft train on data sent to the Phi-4 API, and is commercial use allowed?+
100% Isolated when self-hosted on private GPUs or VPC. License tier: Open-Weights Research & Commercial License. Full commercial ownership of generated outputs. Supported deployment channels include: Self-Hosted GPU (vLLM / Ollama), Private VPC, or Managed Serverless API.
How does Phi-4 perform on coding and reasoning benchmarks?+
Phi-4 achieves a composite Intelligence Index of 80/100, scoring N/A% on SWE-bench Verified, 55% on GPQA Diamond, 78.5% on MMLU-Pro, and 1305 pts on Chatbot Arena Elo.
What is Phi-4 best used for?+
Phi-4 is best suited for: local perimetral desarrollador asistentes, offline matemáticas problem-solving, y efficient workstation tareas. Assessment: de Microsoft masterclass en pequeño modelo de lenguaje eficiencia, proving que datos calidad triumphs sobre escala.
Microsoft Family Lineage, Routing Tiers & Upgrade Path
Step down to a lighter sibling for high-volume routing or step up to Microsoft's flagship reasoning tier:
AI Glossary & Architecture Concepts
View full glossary →Key machine learning concepts and benchmarks relevant to Phi-4:
La unidad básica de texto que procesa un modelo de lenguaje, que normalmente equivale a 3 o 4 caracteres o aproximadamente 0,75 palabras. Por ejemplo, una frase típica se divide en varios tokens.
Un conjunto estandarizado de pruebas para evaluar y comparar objetivamente las capacidades de distintos modelos de IA en tareas como programación (SWE-bench) o matemáticas avanzadas (AIME).
La cantidad máxima de tokens que un modelo puede considerar a la vez, incluyendo tanto la entrada que lee como la salida que genera. Una ventana de 1.000.000 de tokens puede procesar unas 750.000 palabras en una sola petición.
Un modelo de IA entrenado para invertir cómputo adicional en pensar y validar hipótesis lógicas antes de responder, diseñado para problemas complejos de ingeniería y matemáticas.
Una técnica de razonamiento paso a paso donde el modelo genera pensamientos intermedios antes de formular la respuesta final, mejorando radicalmente la resolución de problemas matemáticos y lógicos.
La capacidad de un modelo de detectar la necesidad de una función externa y emitir argumentos JSON estructurados para ejecutar API, búsquedas web o consultas a bases de datos.
Las puntuaciones de benchmarks, precios por token y límites de tasa evolucionan rápidamente. Verifique siempre las especificaciones actuales en la documentación oficial del proveedor.
Fuentes: Microsoft Research Phi-4 Technical Report & Hugging Face↗