Phi-3.5 7B: pricing, context window and benchmarks
Phi-3.5 7B is a small open-weight model from Microsoft. It offers a 4,096-token context window, costs $0.50 per million input tokens and $2.68 per million output tokens, and averages 50.5 across the index's benchmark suite.
Pricing
Phi-3.5 7B costs $0.50 per million input tokens and $2.68 per million output tokens. A blended 3:1 input-to-output mix works out to about $1.04 per million tokens. Cached input is listed at $0.07 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $0.50 | per 1M tokens |
| Output | $2.68 | per 1M tokens |
| Cached input | $0.07 | per 1M tokens |
| Blended 3:1 | $1.04 | estimate per 1M tokens |
| 1M in + 1M out | $3.18 | 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 | 59.3 | 534 | 11th | |
| GPQA Diamond | 47.6 | 542 | 10th | |
| SWE-bench Verified | 33.2 | 570 | 5th | |
| HumanEval | 54.5 | 547 | 9th | |
| MATH-500 | 51.8 | 544 | 9th | |
| AIME 2025 | 36.8 | 557 | 7th | |
| LiveCodeBench | 35.2 | 569 | 5th | |
| IFEval | 61.5 | 561 | 7th | |
| TruthfulQA | 55.0 | 574 | 4th | |
| HellaSwag | 69.1 | 546 | 9th | |
| MMMU | 51.1 | 543 | 10th | |
| Arena Elo | 1190.4 | 565 | 6th |
What Phi-3.5 7B is good for
Phi-3.5 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 $0.50 per million input tokens it can handle a large share of production traffic.
Strengths
- Input price of $0.50 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 4,096 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 |
|---|---|---|---|---|
| Nemotron 4 1.5B | NVIDIA | 32768 | $0.49 | 50.8 |
| Llama 3.1 1.5B | Meta AI | 4096 | $0.53 | 50.7 |
| Hunyuan-Large 1.5B | Tencent | 8192 | $0.54 | 50.5 |
| Phi-3.5 7B | Microsoft | 4096 | $0.50 | 50.5 |
| Arctic 7B | Snowflake | 8192 | $0.49 | 50.3 |
| Llama 3.1 Cerebras 1.5B | Cerebras | 4096 | $0.48 | 50.2 |
FAQ
How much does Phi-3.5 7B cost?
Phi-3.5 7B is listed at $0.50 per million input tokens and $2.68 per million output tokens. Cached input is $0.072 per million tokens. A blended 3:1 input-to-output mix is about $1.04 per million tokens.
What is the context window of Phi-3.5 7B?
Phi-3.5 7B has a context window of 4,096 tokens and can emit up to 4,096 output tokens per request in this index.
Is Phi-3.5 7B open weights?
Yes. Phi-3.5 7B is listed as an open-weight model under a open weights licence.
How does Phi-3.5 7B score on benchmarks?
Phi-3.5 7B averages 50.5 across the scored benchmarks in this index, ranking #551 of 600 models. Its strongest relative result is Arena Elo at 1190.4.
What is Phi-3.5 7B best for?
Phi-3.5 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 $0.50 per million input tokens it can handle a large share of production traffic.