DeepSeek V3.2 (ByteDance) vs DeepSeek V4 Flash (reka-ai)
DeepSeek V3.2 (ByteDance) is cheaper on output tokens, while DeepSeek V4 Flash (reka-ai) offers a larger context window. Choose DeepSeek V3.2 (ByteDance) or DeepSeek V4 Flash (reka-ai) based on the trade-off between cost, context, and the benchmarks that matter for your use case.
| Spec | DeepSeek V3.2 (ByteDance) | DeepSeek V4 Flash (reka-ai) |
|---|---|---|
| Provider | LLM Gateway | LLM Gateway |
| Input / 1M tokens | $0.28 | $0.11 |
| Output / 1M tokens | $0.42 | $0.66 |
| Context window | 131K | 262K |
| Parameters | — | — |
| Open weights | Yes | Yes |
| Released | Dec 2025 | Apr 2026 |
FAQ
Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.