Ministral 8B (latest) vs DeepSeek V4 Flash 0731 (Thinking)
DeepSeek V4 Flash 0731 (Thinking) is cheaper on output tokens, while DeepSeek V4 Flash 0731 (Thinking) offers a larger context window. Choose Ministral 8B (latest) or DeepSeek V4 Flash 0731 (Thinking) based on the trade-off between cost, context, and the benchmarks that matter for your use case.
| Spec | Ministral 8B (latest) | DeepSeek V4 Flash 0731 (Thinking) |
|---|---|---|
| Provider | Mistral | NanoGPT |
| Input / 1M tokens | $0.15 | $0.04 |
| Output / 1M tokens | $0.15 | $0.13 |
| Context window | 262K | 1M |
| Parameters | — | — |
| Open weights | Yes | Yes |
| Released | Oct 2024 | Jul 2026 |
| GPQA Diamond score | 27.1 | — |
| LMArena (Chatbot Arena) Elo score | 1191 | — |
FAQ
Is Ministral 8B (latest) or DeepSeek V4 Flash 0731 (Thinking) cheaper?
DeepSeek V4 Flash 0731 (Thinking) is cheaper on output tokens ($0.13 vs $0.15 per 1M).
Which has the larger context window, Ministral 8B (latest) or DeepSeek V4 Flash 0731 (Thinking)?
DeepSeek V4 Flash 0731 (Thinking) has the larger context window (1M tokens).
Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.