Phi-4 7B: pricing, context window and benchmarks
Phi-4 7B is a small open-weight model from Microsoft. It offers a 8,192-token context window, costs $0.91 per million input tokens and $3.55 per million output tokens, and averages 56.8 across the index's benchmark suite.
Pricing
Phi-4 7B costs $0.91 per million input tokens and $3.55 per million output tokens. A blended 3:1 input-to-output mix works out to about $1.57 per million tokens. Cached input is listed at $0.13 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $0.91 | per 1M tokens |
| Output | $3.55 | per 1M tokens |
| Cached input | $0.13 | per 1M tokens |
| Blended 3:1 | $1.57 | estimate per 1M tokens |
| 1M in + 1M out | $4.46 | estimate |
Benchmarks
Scores below are modelled estimates on a 0–100 scale. Use them to compare relative position, not as official vendor results.
| Benchmark | Score | Rank | Percentile | Relative |
|---|---|---|---|---|
| MMLU-Pro | 62.4 | 498 | 17th | |
| GPQA Diamond | 52.7 | 492 | 18th | |
| SWE-bench Verified | 44.0 | 479 | 20th | |
| HumanEval | 61.3 | 479 | 20th | |
| MATH-500 | 55.6 | 514 | 14th | |
| AIME 2025 | 48.7 | 465 | 23th | |
| LiveCodeBench | 43.3 | 514 | 14th | |
| IFEval | 67.9 | 491 | 18th | |
| TruthfulQA | 60.2 | 527 | 12th | |
| HellaSwag | 71.4 | 514 | 14th | |
| MMMU | 57.1 | 469 | 22th | |
| Arena Elo | 1241.9 | 508 | 15th |
What Phi-4 7B is good for
Phi-4 7B is a small model. That makes it a good fit for classification, extraction, routing, summarization and other high-volume jobs where latency and cost dominate. It is not the right pick for open-ended frontier reasoning, but at $0.91 per million input tokens it can handle a large share of production traffic.
Strengths
- Input price of $0.91 per million tokens is at or below the index median.
- Open weights make self-hosting, fine-tuning and data-residency review possible.
Trade-offs
- Context window of 8,192 tokens is small for long-document work.
- Average benchmark score is below the index median; test it on your own task.
- Less reliable on open-ended reasoning and complex multi-step instructions.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| DeepSeek-V3.2 1.5B | DeepSeek | 4096 | $1.02 | 56.9 |
| Nemotron 4 7B | NVIDIA | 16384 | $1.05 | 56.9 |
| Fusion Nano | OpenRouter | 8192 | $2.96 | 56.8 |
| Kimi K2 1.5B | Moonshot AI | 4096 | $0.99 | 56.8 |
| Phi-4 7B | Microsoft | 8192 | $0.91 | 56.8 |
| Llama 3.3 Groq 7B | Groq | 16384 | $1.01 | 56.7 |
FAQ
How much does Phi-4 7B cost?
Phi-4 7B is listed at $0.91 per million input tokens and $3.55 per million output tokens. Cached input is $0.13 per million tokens. A blended 3:1 input-to-output mix is about $1.57 per million tokens.
What is the context window of Phi-4 7B?
Phi-4 7B has a context window of 8,192 tokens and can emit up to 8,192 output tokens per request in this index.
Is Phi-4 7B open weights?
Yes. Phi-4 7B is listed as an open-weight model under a open weights licence.
How does Phi-4 7B score on benchmarks?
Phi-4 7B averages 56.8 across the scored benchmarks in this index, ranking #494 of 600 models. Its strongest relative result is Arena Elo at 1241.9.
What is Phi-4 7B best for?
Phi-4 7B is a small model. That makes it a good fit for classification, extraction, routing, summarization and other high-volume jobs where latency and cost dominate. It is not the right pick for open-ended frontier reasoning, but at $0.91 per million input tokens it can handle a large share of production traffic.