Phi 4: Benchmarks, Pricing & Context Window
Phi 4 is a language model from Microsoft, released in December 2024, with a 16K-token context window, and pricing from $0.070/M input and $0.140/M output.
phi-4 is a state-of-the-art open model built to excel at advanced reasoning, coding, and knowledge tasks. It leverages a blend of synthetic data, filtered web data, academic texts, and supervised fine-tuning for precision, alignment, and
Phi 4 benchmarks
Capability tiers
Standing within each category, adjusted for leaderboard depth.
Real tasks performance
High-confidence performance for Phi 4 across real-world prompt categories. Only 95% intervals at most 4 points wide are shown.
Performance by conversation depth
How Phi 4 holds up as conversations get longer.
Quality Tracker
Phi 4 Performance Across Datasets
Scores sourced from the model's scorecard, paper, or official blog posts
llm-stats.com - Fri Sep 18 2026Phi 4 pricing
Providers
Phi 4 starts at $0.0700 per million input tokens and $0.140 per million output tokens via DeepInfra.
| Provider | Input $/M | Cached input $/M | Output $/M | Context in / out | TTFT p95 s | Output p5 c/s | Modalities in / out |
|---|---|---|---|---|---|---|---|
 DeepInfra | $0.0700 | — | $0.140 | 16.4K/16.4K | — | — | / |
Cached input is the discounted price for prompt tokens served from a provider cache. TTFT is time to first token. Output is characters per second; p5 is the sustained floor exceeded by 95% of observed requests.
Phi 4 model size
Phi 4 has 14.7 billion parameters and was trained on 9.8 trillion tokens. See how it compares to other models in the same parameter range.
Phi 4 context window
Input and output token limits for Phi 4, plus how it ranks on long-context understanding.
Try now
huggleMake it with
Phi 4.
Phi 4 latency
Phi 4 time to first token, sustained output throughput, and failed-request rate from live model usage over the trailing 7 days.
Phi 4 examples
Recent arena outputs from Phi 4, picked from the highest-ranked matchups.
Phi 4 license
Phi 4 is released under the MIT license, which permits commercial use, has 14.7B parameters, has a knowledge cutoff of June 2024.
- License
- MIT
- Commercial use allowed
- Parameters
- 14.7B
- Knowledge cutoff
- June 2024
MIT License - allows commercial use
Phi 4 resources
Official sources for Phi 4: provider documentation, paper or system card, official launch post.
Phi 4 vs other models
The most-compared alternatives to Phi 4 are Grok-2, Llama 3.1 70B Instruct, Claude 3.5 Sonnet. Open any pair side-by-side for benchmarks, pricing, context, and latency.
Models like Phi 4
Models ranked just above and below Phi 4 by LLM Stats score.
FAQ
Common questions about Phi 4.