Phi-3.5 Base: pricing, context window and benchmarks
Phi-3.5 Base is a open-weight model from Microsoft. It offers a 65,536-token context window, costs $2.24 per million input tokens and $10.25 per million output tokens, and averages 66.6 across the index's benchmark suite.
Pricing
Phi-3.5 Base costs $2.24 per million input tokens and $10.25 per million output tokens. A blended 3:1 input-to-output mix works out to about $4.24 per million tokens. Cached input is listed at $0.28 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $2.24 | per 1M tokens |
| Output | $10.25 | per 1M tokens |
| Cached input | $0.28 | per 1M tokens |
| Blended 3:1 | $4.24 | estimate per 1M tokens |
| 1M in + 1M out | $12.49 | 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 | 72.4 | 379 | 37th | |
| GPQA Diamond | 60.3 | 402 | 33th | |
| SWE-bench Verified | 51.6 | 409 | 32th | |
| HumanEval | 73.1 | 371 | 38th | |
| MATH-500 | 71.6 | 373 | 38th | |
| AIME 2025 | 59.7 | 389 | 35th | |
| LiveCodeBench | 55.2 | 401 | 33th | |
| IFEval | 74.2 | 404 | 33th | |
| TruthfulQA | 73.0 | 370 | 38th | |
| HellaSwag | 79.5 | 389 | 35th | |
| MMMU | 61.7 | 412 | 31th | |
| Arena Elo | 1315.0 | 400 | 33th |
What Phi-3.5 Base is good for
Phi-3.5 Base is an open-weight model, so it can be self-hosted or run through a third-party inference provider. That matters for data residency, fine-tuning and long-run cost control. At $2.24 per million input tokens (hosted list price) and a 65,536-token context window, it is a practical option for teams that want more control than a closed API allows.
Strengths
- Input price of $2.24 per million tokens is at or below the index median.
- Open weights make self-hosting, fine-tuning and data-residency review possible.
- Accepts more than text, including image input where the provider supports it.
Trade-offs
- Average benchmark score is below the index median; test it on your own task.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Molmo Chat | Allen Institute for AI | 32768 | $2.40 | 66.7 |
| Doubao 1.5 Pro Nano | ByteDance | 16384 | $5.78 | 66.6 |
| Nemotron Nano | NVIDIA | 8192 | $2.24 | 66.6 |
| Phi-3.5 Base | Microsoft | 65536 | $2.24 | 66.6 |
| Command A Nano | Cohere | 8192 | $5.90 | 66.3 |
| Nemotron Nano Chat | NVIDIA | 32768 | $2.21 | 66.2 |
FAQ
How much does Phi-3.5 Base cost?
Phi-3.5 Base is listed at $2.24 per million input tokens and $10.25 per million output tokens. Cached input is $0.28 per million tokens. A blended 3:1 input-to-output mix is about $4.24 per million tokens.
What is the context window of Phi-3.5 Base?
Phi-3.5 Base has a context window of 65,536 tokens and can emit up to 16,384 output tokens per request in this index.
Is Phi-3.5 Base open weights?
Yes. Phi-3.5 Base is listed as an open-weight model under a open weights licence.
How does Phi-3.5 Base score on benchmarks?
Phi-3.5 Base averages 66.6 across the scored benchmarks in this index, ranking #391 of 600 models. Its strongest relative result is Arena Elo at 1315.0.
What is Phi-3.5 Base best for?
Phi-3.5 Base is an open-weight model, so it can be self-hosted or run through a third-party inference provider. That matters for data residency, fine-tuning and long-run cost control. At $2.24 per million input tokens (hosted list price) and a 65,536-token context window, it is a practical option for teams that want more control than a closed API allows.