Head-to-head正面对比
DeepSeek V4 Flash 0423 vs Qwen3.7 Plus
DeepSeek meets Qwen: sourced benchmark scores side by side, dimension by dimension, with prices and specs. Every conclusion below is derived from the published dataset — no opinions, no sponsorship.
DeepSeek 对 Qwen:带来源的 benchmark 成绩逐项并排,附价格与规格。下面每条结论都从公开数据集推导——没有主观意见,没有厂商赞助。
◈
The verdict, derived from data结论(全部由数据推导)
- On overall score they are nearly tied (82 vs 81.7) — decide on price, context and workload mix.总分几乎打平(82 对 81.7)——按价格、上下文和任务结构来选。
- DeepSeek V4 Flash 0423 is stronger in math.DeepSeek V4 Flash 0423 在数学上更强。
- Qwen3.7 Plus is stronger in preference.Qwen3.7 Plus 在偏好上更强。
- DeepSeek V4 Flash 0423 is 8.1× cheaper on blended price ($0.099 vs $0.800 per 1M tokens, in/out average).混合价(输入输出均值)上 DeepSeek V4 Flash 0423 便宜 8.1 倍:$0.099 对 $0.800/百万 token。
● DeepSeek V4 Flash 0423 ● Qwen3.7 Plus · missing dimensions count as 0 in the chart缺数据的维度在图上按 0 计
≡
Spec by spec逐项规格
| DeepSeek V4 Flash 0423 | Qwen3.7 Plus | |
|---|---|---|
| Provider厂商 | DeepSeek | Qwen |
| Released发布 | 2026-04-24 | 2026-06-03 |
| Intelligence Score智能评分 | 82 | 81.7 |
| Benchmark coveragebenchmark 覆盖 | 88% | 76% |
| Reasoning推理 | 73.7 | 77.5 |
| Coding代码 | 88.4 | 86.9 |
| Knowledge知识 | 90.7 | 93.2 |
| Math数学 | 97 | 85.7 |
| Agent智能体 | 96.5 | 93.8 |
| Preference偏好 | 34 | 39.8 |
| API input $/1M输入价 $/百万 | $0.066 | $0.320 |
| API output $/1M输出价 $/百万 | $0.131 | $1.28 |
| Blended $/1M混合价 $/百万 | $0.099 | $0.800 |
| Value (score per $)性价比(分数/美元) | 832.5 | 102.1 |
| Context window上下文 | 1M | 1M |
| Max output最大输出 | 384K | 131K |
| Reasoning tiers推理档位 | thinking off / high effort / max effort | thinking on / off |
| Open weights开放权重 | Yes支持 | Yes支持 |
| CN-direct国内直连 | Yes支持 | Yes支持 |
| Free tier免费层 | No否 | No否 |
⚔
Benchmark by benchmark逐项 benchmark
| Benchmark基准 | DeepSeek V4 Flash 0423 | Qwen3.7 Plus |
|---|---|---|
| Humanity's Last Exam | 32.1 | 34.7 |
| GPQA Diamond | 87.4 | 90.3 |
| MMLU-Pro | 86.2 | 88.5 |
| AIME 2025 | 97 | 85.7 |
| LiveCodeBench | 91.6⚡max | 89.6 |
| Terminal-Bench | 82.7 | 70.3 |
| τ²-bench | 95.6⚡high | 93 |
| LMArena (Chatbot Arena) | 1435 | 1458 |
DeepSeek V4 Flash 0423 dossier → 档案 → · Qwen3.7 Plus dossier → 档案 → · How we score评分方法