Phi-3.5 8B: pricing, context window and benchmarks
Phi-3.5 8B is a small open-weight model from Microsoft. It offers a 16,384-token context window, costs $0.59 per million input tokens and $2.75 per million output tokens, and averages 51.2 across the index's benchmark suite.
Pricing
Phi-3.5 8B costs $0.59 per million input tokens and $2.75 per million output tokens. A blended 3:1 input-to-output mix works out to about $1.13 per million tokens. Cached input is listed at $0.05 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $0.59 | per 1M tokens |
| Output | $2.75 | per 1M tokens |
| Cached input | $0.05 | per 1M tokens |
| Blended 3:1 | $1.13 | estimate per 1M tokens |
| 1M in + 1M out | $3.34 | 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 | 58.6 | 544 | 9th | |
| GPQA Diamond | 47.2 | 546 | 9th | |
| SWE-bench Verified | 38.8 | 530 | 12th | |
| HumanEval | 55.5 | 539 | 10th | |
| MATH-500 | 52.3 | 538 | 10th | |
| AIME 2025 | 37.8 | 553 | 8th | |
| LiveCodeBench | 40.7 | 534 | 11th | |
| IFEval | 61.9 | 555 | 8th | |
| TruthfulQA | 56.6 | 559 | 7th | |
| HellaSwag | 66.6 | 571 | 5th | |
| MMMU | 47.3 | 570 | 5th | |
| Arena Elo | 1200.9 | 553 | 8th |
What Phi-3.5 8B is good for
Phi-3.5 8B 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 $0.59 per million input tokens it can handle a large share of production traffic.
Strengths
- Input price of $0.59 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 16,384 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 |
|---|---|---|---|---|
| OLMo 2 7B | Allen Institute for AI | 16384 | $0.56 | 51.7 |
| QwQ 1.5B | Alibaba Cloud | 32768 | $0.61 | 51.6 |
| MiniMax-Text 1.5B | MiniMax | 32768 | $0.58 | 51.5 |
| Phi-3.5 8B | Microsoft | 16384 | $0.59 | 51.2 |
| DeepSeek-Coder 1.5B | DeepSeek | 16384 | $0.58 | 51.0 |
| Nemotron 4 1.5B | NVIDIA | 32768 | $0.49 | 50.8 |
FAQ
How much does Phi-3.5 8B cost?
Phi-3.5 8B is listed at $0.59 per million input tokens and $2.75 per million output tokens. Cached input is $0.049 per million tokens. A blended 3:1 input-to-output mix is about $1.13 per million tokens.
What is the context window of Phi-3.5 8B?
Phi-3.5 8B has a context window of 16,384 tokens and can emit up to 4,096 output tokens per request in this index.
Is Phi-3.5 8B open weights?
Yes. Phi-3.5 8B is listed as an open-weight model under a open weights licence.
How does Phi-3.5 8B score on benchmarks?
Phi-3.5 8B averages 51.2 across the scored benchmarks in this index, ranking #546 of 600 models. Its strongest relative result is Arena Elo at 1200.9.
What is Phi-3.5 8B best for?
Phi-3.5 8B 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 $0.59 per million input tokens it can handle a large share of production traffic.