Model comparison
DeepSeek V4 Flash vs Qwen3.8 Max
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 53.6 on the Noometry Index. DeepSeek V4 Flash costs 11× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Last verified . 32 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 32 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 0 categories and Qwen3.8 Max in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3.8 Max leads 73.2 to 60.3.
- The biggest single-benchmark swing is FrontierMath Tier 4: 24.4% for DeepSeek V4 Flash and 46.3% for Qwen3.8 Max.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $2 / $6 for Qwen3.8 Max.
- DeepSeek V4 Flash has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4 Flash | Qwen3.8 Max | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 53.6 | 56.8 |
| Released | 2026-04-24 | 2026-08-02 |
| Weights | Open | Proprietary |
| Context window | 1M | 1M |
| Max output | 393K | 131K |
| Input $ / M tokens | $0.15 | $2 |
| Output $ / M tokens | $0.60 | $6 |
| Results tracked | 41 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 Max leads
DeepSeek V4 Flash: 47.9 (#59), Qwen3.8 Max: 53.5 (#29)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| LMArena WebDev | 1582 | 1674 |
| SciCode | 49.9% | 53.2% |
| LMArena Coding | 1457 | 1502 |
| DeepSWE | — | 57.5% |
| FrontierCode | 18.8% | — |
| FrontierSWE | — | 17.8% |
| WeirdML | 63% | — |
| ALE-Bench | 1,306 | — |
Agentic & Tool Use Not comparable
DeepSeek V4 Flash: —, Qwen3.8 Max: 45.4 (#14)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| APEX-Agents | — | 63.3% |
| τ²-bench Banking | — | 55.1% |
| GDP.pdf | — | 23.2% |
Reasoning Too close to call
DeepSeek V4 Flash: 53.7 (#30), Qwen3.8 Max: 54.4 (#26)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| NYT Connections (extended) | 89.6% | 88.3% |
| CritPt | 16.6% | 20% |
| Chess Puzzles | 33% | 40% |
| LMArena Hard Prompts | 1444 | 1496 |
| Mystery Game Puzzles | 34% | 38% |
| DTBench | 90.9% | 92% |
| LMCA | 41.7% | 46.2% |
| Epoch Capabilities Index | 154.49 | 156.41 |
| ARC-AGI-2 | 61.4% | — |
| SimpleBench | 61.1% | — |
| Kagi LLM Benchmark | 52.2% | — |
| ARC-AGI-1 | 89% | — |
Math Qwen3.8 Max leads
DeepSeek V4 Flash: 60.3 (#37), Qwen3.8 Max: 73.2 (#20)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| FrontierMath (Tiers 1-3) | 57.5% | 74.7% |
| FrontierMath Tier 4 | 24.4% | 46.3% |
| OTIS Mock AIME 2024-2025 | 94.4% | 100% |
| ProofBench | 56% | 58% |
| LMArena Math | 1427 | 1499 |
| MathArena Final-Answer Competitions | 76.5% | — |
Knowledge Qwen3.8 Max leads
DeepSeek V4 Flash: 55.4 (#48), Qwen3.8 Max: 61.7 (#27)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | 91% | 92.7% |
| SimpleQA Verified | 33.6% | 47.3% |
| LMArena Expert | 1441 | 1507 |
Multimodal Not comparable
DeepSeek V4 Flash: —, Qwen3.8 Max: 37.2 (#75)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | — | 1314 |
| Furniture Assembly | — | 20% |
Multilingual Qwen3.8 Max leads
DeepSeek V4 Flash: 53.0 (#72), Qwen3.8 Max: 56.7 (#18)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | 1420 | 1472 |
| LMArena Chinese | 1468 | 1538 |
| LMArena French | 1439 | 1503 |
| LMArena German | 1418 | 1483 |
| LMArena Japanese | 1406 | 1467 |
| LMArena Korean | 1384 | 1461 |
| LMArena Russian | 1428 | 1481 |
| LMArena Spanish | 1436 | 1492 |
Instruction Following Qwen3.8 Max leads
DeepSeek V4 Flash: 74.9 (#81), Qwen3.8 Max: 77.6 (#17)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | 1421 | 1479 |
Long Context Qwen3.8 Max leads
DeepSeek V4 Flash: 43.8 (#85), Qwen3.8 Max: 45.6 (#31)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | 1434 | 1489 |
Writing & Preference Qwen3.8 Max leads
DeepSeek V4 Flash: 63.8 (#61), Qwen3.8 Max: 67.1 (#30)
| Benchmark | DeepSeek V4 Flash | Qwen3.8 Max |
|---|---|---|
| LMArena Text | 1432 | 1483 |
| LMArena Creative Writing | 1403 | 1479 |
| LMArena Multi-Turn | 1449 | 1489 |
| EQ-Bench Creative Writing | 1559 | — |
Frequently asked questions
Is DeepSeek V4 Flash better than Qwen3.8 Max?
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 53.6 on the Noometry Index. DeepSeek V4 Flash costs 11× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Which is cheaper, DeepSeek V4 Flash or Qwen3.8 Max?
DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Qwen3.8 Max lists at $2 and $6.
Is DeepSeek V4 Flash or Qwen3.8 Max better for coding?
Qwen3.8 Max scores higher on coding benchmarks: 53.5 versus 47.9 in the Noometry coding category.
Which has the bigger context window?
Both accept 1M tokens.
How many benchmarks do DeepSeek V4 Flash and Qwen3.8 Max share?
32 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Qwen3.8 Max has 39.