Sonar Pro: pricing, context window and benchmarks
Sonar Pro is a multimodal model from Perplexity. It offers a 128,000-token context window, costs $13.01 per million input tokens and $50.93 per million output tokens, and averages 79.1 across the index's benchmark suite.
Pricing
Sonar Pro costs $13.01 per million input tokens and $50.93 per million output tokens. A blended 3:1 input-to-output mix works out to about $22.49 per million tokens. Cached input is listed at $1.58 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $13.01 | per 1M tokens |
| Output | $50.93 | per 1M tokens |
| Cached input | $1.58 | per 1M tokens |
| Blended 3:1 | $22.49 | estimate per 1M tokens |
| 1M in + 1M out | $63.94 | 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 | 78.7 | 224 | 63th | |
| GPQA Diamond | 77.1 | 105 | 83th | |
| SWE-bench Verified | 68.9 | 132 | 78th | |
| HumanEval | 82.6 | 170 | 72th | |
| MATH-500 | 80.7 | 196 | 67th | |
| AIME 2025 | 74.7 | 162 | 73th | |
| LiveCodeBench | 75.0 | 96 | 84th | |
| IFEval | 87.4 | 125 | 79th | |
| TruthfulQA | 82.5 | 119 | 80th | |
| HellaSwag | 86.5 | 196 | 67th | |
| MMMU | 75.7 | 146 | 76th | |
| Arena Elo | 1411.3 | 181 | 70th |
What Sonar Pro is good for
Sonar Pro is tuned toward coding work. It is a reasonable candidate for code generation, review, debugging and repository-scale tasks, especially when paired with a retrieval layer. At $13.01 per million input tokens, it sits in the mid-range of the index, and the 128,000-token context window is enough for large files and multi-file diffs.
Strengths
- Large 128,000-token context window for long documents and agent traces.
- Accepts more than text, including image input where the provider supports it.
- Coding benchmarks are a relative strength within its tier.
Trade-offs
- Proprietary licence: you cannot download the weights or self-host the model.
- Premium input pricing at $13.01 per million tokens.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Llama 4 Maverick Base | Meta AI | 128000 | $4.68 | 79.1 |
| Llama 4 Scout Instruct | Meta AI | 32768 | $4.59 | 79.1 |
| Mistral Large 3 Chat | Mistral AI | 1000000 | $4.90 | 79.1 |
| Qwen3 Chat | Alibaba Cloud | 128000 | $4.75 | 79.1 |
| Sonar Pro | Perplexity | 128000 | $13.01 | 79.1 |
| Claude Opus 4.5 Mini | Anthropic | 65536 | $12.72 | 79.0 |
FAQ
How much does Sonar Pro cost?
Sonar Pro is listed at $13.01 per million input tokens and $50.93 per million output tokens. Cached input is $1.58 per million tokens. A blended 3:1 input-to-output mix is about $22.49 per million tokens.
What is the context window of Sonar Pro?
Sonar Pro has a context window of 128,000 tokens and can emit up to 32,768 output tokens per request in this index.
Is Sonar Pro open weights?
No. Sonar Pro is a proprietary model with a proprietary licence, so you cannot download the weights.
How does Sonar Pro score on benchmarks?
Sonar Pro averages 79.1 across the scored benchmarks in this index, ranking #148 of 600 models. Its strongest relative result is Arena Elo at 1411.3.
What is Sonar Pro best for?
Sonar Pro is tuned toward coding work. It is a reasonable candidate for code generation, review, debugging and repository-scale tasks, especially when paired with a retrieval layer. At $13.01 per million input tokens, it sits in the mid-range of the index, and the 128,000-token context window is enough for large files and multi-file diffs.