Phi-4 1.5B: pricing, context window and benchmarks
Phi-4 1.5B is a small open-weight model from Microsoft. It offers a 8,192-token context window, costs $0.42 per million input tokens and $1.90 per million output tokens, and averages 48.1 across the index's benchmark suite.
Pricing
Phi-4 1.5B costs $0.42 per million input tokens and $1.90 per million output tokens. A blended 3:1 input-to-output mix works out to about $0.79 per million tokens. Cached input is listed at $0.05 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $0.42 | per 1M tokens |
| Output | $1.90 | per 1M tokens |
| Cached input | $0.05 | per 1M tokens |
| Blended 3:1 | $0.79 | estimate per 1M tokens |
| 1M in + 1M out | $2.32 | 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 | 55.0 | 573 | 5th | |
| GPQA Diamond | 46.3 | 556 | 7th | |
| SWE-bench Verified | 35.2 | 556 | 7th | |
| HumanEval | 50.9 | 569 | 5th | |
| MATH-500 | 47.6 | 571 | 5th | |
| AIME 2025 | 32.8 | 578 | 4th | |
| LiveCodeBench | 36.9 | 558 | 7th | |
| IFEval | 58.9 | 578 | 4th | |
| TruthfulQA | 53.5 | 582 | 3th | |
| HellaSwag | 65.7 | 579 | 4th | |
| MMMU | 46.1 | 580 | 3th | |
| Arena Elo | 1181.3 | 574 | 4th |
What Phi-4 1.5B is good for
Phi-4 1.5B 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 $0.42 per million input tokens it can handle a large share of production traffic.
Strengths
- Input price of $0.42 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 |
|---|---|---|---|---|
| Llama 3 Hyperbolic 1.5B | Hyperbolic | 32768 | $0.45 | 48.4 |
| Llama 3 Lambda 1.5B | Lambda | 16384 | $0.46 | 48.4 |
| Llama 3 Baseten 1.5B | Baseten | 4096 | $0.42 | 48.2 |
| Nemotron Nano 8B | NVIDIA | 8192 | $0.46 | 48.1 |
| Phi-4 1.5B | Microsoft | 8192 | $0.42 | 48.1 |
| Mixtral 1.5B | Mistral AI | 16384 | $0.36 | 47.7 |
FAQ
How much does Phi-4 1.5B cost?
Phi-4 1.5B is listed at $0.42 per million input tokens and $1.90 per million output tokens. Cached input is $0.047 per million tokens. A blended 3:1 input-to-output mix is about $0.79 per million tokens.
What is the context window of Phi-4 1.5B?
Phi-4 1.5B 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 1.5B open weights?
Yes. Phi-4 1.5B is listed as an open-weight model under a open weights licence.
How does Phi-4 1.5B score on benchmarks?
Phi-4 1.5B averages 48.1 across the scored benchmarks in this index, ranking #574 of 600 models. Its strongest relative result is Arena Elo at 1181.3.
What is Phi-4 1.5B best for?
Phi-4 1.5B 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 $0.42 per million input tokens it can handle a large share of production traffic.