DeepSeek-V3.2 8B: pricing, context window and benchmarks
DeepSeek-V3.2 8B is a small open-weight model from DeepSeek. It offers a 32,768-token context window, costs $2.02 per million input tokens and $9.37 per million output tokens, and averages 66.2 across the index's benchmark suite.
Pricing
DeepSeek-V3.2 8B costs $2.02 per million input tokens and $9.37 per million output tokens. A blended 3:1 input-to-output mix works out to about $3.86 per million tokens. Cached input is listed at $0.17 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $2.02 | per 1M tokens |
| Output | $9.37 | per 1M tokens |
| Cached input | $0.17 | per 1M tokens |
| Blended 3:1 | $3.86 | estimate per 1M tokens |
| 1M in + 1M out | $11.39 | 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 | 378 | 37th | |
| GPQA Diamond | 58.8 | 419 | 30th | |
| SWE-bench Verified | 55.2 | 382 | 36th | |
| HumanEval | 69.7 | 398 | 34th | |
| MATH-500 | 66.7 | 410 | 32th | |
| AIME 2025 | 56.4 | 409 | 32th | |
| LiveCodeBench | 58.9 | 379 | 37th | |
| IFEval | 76.0 | 388 | 35th | |
| TruthfulQA | 71.6 | 386 | 36th | |
| HellaSwag | 77.4 | 409 | 32th | |
| MMMU | 65.4 | 377 | 37th | |
| Arena Elo | 1308.4 | 410 | 32th |
What DeepSeek-V3.2 8B is good for
DeepSeek-V3.2 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 $2.02 per million input tokens it can handle a large share of production traffic.
Strengths
- Input price of $2.02 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.
- Less reliable on open-ended reasoning and complex multi-step instructions.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Nemotron Nano | NVIDIA | 8192 | $2.24 | 66.6 |
| Command A Nano | Cohere | 8192 | $5.90 | 66.3 |
| DeepSeek-V3.2 8B | DeepSeek | 32768 | $2.02 | 66.2 |
| Nemotron Nano Chat | NVIDIA | 32768 | $2.21 | 66.2 |
| Embed 4 Max | Cohere | 8192 | $5.68 | 66.1 |
| Sonar Pro Nano | Perplexity | 32768 | $5.50 | 66.1 |
FAQ
How much does DeepSeek-V3.2 8B cost?
DeepSeek-V3.2 8B is listed at $2.02 per million input tokens and $9.37 per million output tokens. Cached input is $0.17 per million tokens. A blended 3:1 input-to-output mix is about $3.86 per million tokens.
What is the context window of DeepSeek-V3.2 8B?
DeepSeek-V3.2 8B has a context window of 32,768 tokens and can emit up to 16,384 output tokens per request in this index.
Is DeepSeek-V3.2 8B open weights?
Yes. DeepSeek-V3.2 8B is listed as an open-weight model under a open weights licence.
How does DeepSeek-V3.2 8B score on benchmarks?
DeepSeek-V3.2 8B averages 66.2 across the scored benchmarks in this index, ranking #393 of 600 models. Its strongest relative result is Arena Elo at 1308.4.
What is DeepSeek-V3.2 8B best for?
DeepSeek-V3.2 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 $2.02 per million input tokens it can handle a large share of production traffic.