DeepSeek-V4-ProvsQwen3.7-Max-Preview
Across 9 shared benchmarks, Qwen3.7-Max-Preview leads overall: DeepSeek-V4-Pro wins 0, Qwen3.7-Max-Preview wins 9, with 0 ties and an average score difference of -21.77.
DeepSeek-V4-Pro
DeepSeek-AI · 2026-08-13 · Reasoning model
Qwen3.7-Max-Preview
阿里巴巴 · 2026-05-20 · Reasoning model
DeepSeek-V4-Pro0 wins(0%)(100%)9 winsQwen3.7-Max-Preview
Benchmark scores
Grouped by capability, sorted by largest gap within each. 9 shared benchmarks.
Coding and Software Engineer
Qwen3.7-Max-Preview 4/4| Benchmark | DeepSeek-V4-Pro | Qwen3.7-Max-Preview | Diff |
|---|---|---|---|
| LiveCodeBench | 56.8076 / 123Normal (No Tools) | 91.604 / 123最高（无工具） | -34.80 |
| SWE-bench Multilingual | 69.8018 / 23Normal (With Tools) | 78.304 / 23Thinking (With Tools) | -8.50 |
| SWE-Bench Pro - Public | 52.1038 / 55Normal (With Tools) | 60.6010 / 55Thinking (With Tools) | -8.50 |
| SWE-bench Verified | 73.6046 / 113Normal (With Tools) | 80.4013 / 113Thinking (With Tools) | -6.80 |
General Knowledge
Qwen3.7-Max-Preview 2/2| Benchmark | DeepSeek-V4-Pro | Qwen3.7-Max-Preview | Diff |
|---|---|---|---|
| HLE | 7.70159 / 175Normal (No Tools) | 53.5016 / 175Thinking (With Tools) | -45.80 |
| MMLU Pro | 82.9048 / 132Normal (No Tools) | 89.604 / 132最高（无工具） | -6.70 |
AI Agent - Tool Usage
Qwen3.7-Max-Preview 1/1| Benchmark | DeepSeek-V4-Pro | Qwen3.7-Max-Preview | Diff |
|---|---|---|---|
| Terminal Bench 2.0 | 59.1022 / 47Normal (With Tools) | 69.705 / 47Thinking (With Tools) | -10.60 |
General Evaluation
Qwen3.7-Max-Preview 1/1| Benchmark | DeepSeek-V4-Pro | Qwen3.7-Max-Preview | Diff |
|---|---|---|---|
| GPQA Diamond | 72.90146 / 225Normal (No Tools) | 92.4022 / 225最高（无工具） | -19.50 |
Math and Reasoning
Qwen3.7-Max-Preview 1/1| Benchmark | DeepSeek-V4-Pro | Qwen3.7-Max-Preview | Diff |
|---|---|---|---|
| IMO-AnswerBench | 35.3021 / 21Normal (No Tools) | 902 / 21最高（无工具） | -54.70 |
Specs
| Field | DeepSeek-V4-Pro | Qwen3.7-Max-Preview |
|---|---|---|
| Publisher | DeepSeek-AI | 阿里巴巴 |
| Release date | 2026-08-13 | 2026-05-20 |
| Model type | Reasoning model | Reasoning model |
| Architecture | MoE | Dense |
| Parameters | 1.6T | Not available |
| Context length | 1M | 1M |
| Max output | 384K | 64K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | DeepSeek-V4-Pro | Qwen3.7-Max-Preview |
|---|---|---|
| Text input | $0.435 / 1M tokens | $2.5 / 1M tokens |
| Text output | $0.87 / 1M tokens | $7.5 / 1M tokens |
| Cache read | $0.003625 / 1M tokens | $0.25 / 1M tokens |
| Cache write | Not public | $3.125 / 1M tokens |
Summary
- Qwen3.7-Max-Previewleads in:Coding and Software Engineer (4/4), General Knowledge (2/2), AI Agent - Tool Usage (1/1), General Evaluation (1/1), Math and Reasoning (1/1)
On average across the 9 shared benchmarks, Qwen3.7-Max-Preview scores 21.77 higher.
Largest single-benchmark gap: IMO-AnswerBench — DeepSeek-V4-Pro 35.30 vs Qwen3.7-Max-Preview 90 (-54.70).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.