Arctic: pricing, context window and benchmarks
Arctic is a open-weight model from Snowflake. It offers a 8,192-token context window, costs $2.72 per million input tokens and $10.57 per million output tokens, and averages 68.8 across the index's benchmark suite.
Pricing
Arctic costs $2.72 per million input tokens and $10.57 per million output tokens. A blended 3:1 input-to-output mix works out to about $4.68 per million tokens. Cached input is listed at $0.22 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $2.72 | per 1M tokens |
| Output | $10.57 | per 1M tokens |
| Cached input | $0.22 | per 1M tokens |
| Blended 3:1 | $4.68 | estimate per 1M tokens |
| 1M in + 1M out | $13.29 | 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 | 377 | 37th | |
| GPQA Diamond | 65.3 | 354 | 41th | |
| SWE-bench Verified | 54.2 | 391 | 35th | |
| HumanEval | 73.2 | 370 | 38th | |
| MATH-500 | 70.5 | 382 | 36th | |
| AIME 2025 | 64.6 | 343 | 43th | |
| LiveCodeBench | 61.6 | 354 | 41th | |
| IFEval | 79.0 | 354 | 41th | |
| TruthfulQA | 71.5 | 387 | 36th | |
| HellaSwag | 80.5 | 377 | 37th | |
| MMMU | 63.8 | 391 | 35th | |
| Arena Elo | 1345.7 | 370 | 38th |
What Arctic is good for
Arctic 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 $2.72 per million input tokens (hosted list price) and a 8,192-token context window, it is a practical option for teams that want more control than a closed API allows.
Strengths
- Input price of $2.72 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
- 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.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Firesearch Reasoning | Fireworks AI | 32768 | $7.74 | 69.0 |
| Phi-3.5 Chat | Microsoft | 16384 | $2.64 | 69.0 |
| Claude 3.7 Sonnet Nano | Anthropic | 65536 | $6.93 | 68.9 |
| Arctic | Snowflake | 8192 | $2.72 | 68.8 |
| ERNIE 4.0 Mini | Baidu | 32768 | $7.11 | 68.8 |
| Claude Haiku 4.5 Nano | Anthropic | 8192 | $6.47 | 68.7 |
FAQ
How much does Arctic cost?
Arctic is listed at $2.72 per million input tokens and $10.57 per million output tokens. Cached input is $0.22 per million tokens. A blended 3:1 input-to-output mix is about $4.68 per million tokens.
What is the context window of Arctic?
Arctic has a context window of 8,192 tokens and can emit up to 4,096 output tokens per request in this index.
Is Arctic open weights?
Yes. Arctic is listed as an open-weight model under a open weights licence.
How does Arctic score on benchmarks?
Arctic averages 68.8 across the scored benchmarks in this index, ranking #375 of 600 models. Its strongest relative result is Arena Elo at 1345.7.
What is Arctic best for?
Arctic 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 $2.72 per million input tokens (hosted list price) and a 8,192-token context window, it is a practical option for teams that want more control than a closed API allows.