Model comparison
DeepSeek V4.1 Flash vs GPT-5 Pro
DeepSeek V4.1 Flash is the stronger model overall, scoring 52.8 to 46.4 on the Noometry Index.
Last verified . 3 shared benchmarks.
DeepSeekSummary
- They share 3 benchmarks with published results for both. DeepSeek V4.1 Flash scores higher in 4 categories and GPT-5 Pro in 0 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4.1 Flash leads 66.7 to 48.5.
- The biggest single-benchmark swing is FrontierMath (Tiers 1-3): 67.4% for DeepSeek V4.1 Flash and 55.8% for GPT-5 Pro.
- DeepSeek V4.1 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $15 / $120 for GPT-5 Pro.
- DeepSeek V4.1 Flash accepts more context: 1M tokens versus 400K.
- DeepSeek V4.1 Flash has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4.1 Flash | GPT-5 Pro | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 52.8 | 46.4 |
| Released | 2026-09-09 | 2025-10-06 |
| Weights | Open | Proprietary |
| Context window | 1M | 400K |
| Max output | 393K | 272K |
| Input $ / M tokens | $0.15 | $15 |
| Output $ / M tokens | $0.60 | $120 |
| Results tracked | 37 | 12 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4.1 Flash leads
DeepSeek V4.1 Flash: 52.9 (#32), GPT-5 Pro: 44.0 (#80)
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| LMArena WebDev | 1619 | — |
| SciCode | 51.9% | — |
| WeirdML | — | 60.4% |
| LMArena Coding | 1506 | — |
| ALE-Bench | 1,092 | — |
| AlgoTune | — | 1.31 |
Agentic & Tool Use Not comparable
DeepSeek V4.1 Flash: 31.2 (#69), GPT-5 Pro: —
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| APEX-Agents | 39.5% | — |
| GDP.pdf | 19.8% | — |
Reasoning DeepSeek V4.1 Flash leads
DeepSeek V4.1 Flash: 50.2 (#36), GPT-5 Pro: 38.9 (#62)
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| Epoch Capabilities Index | 154.9 | 150.28 |
| ARC-AGI-2 | — | 18.3% |
| SimpleBench | — | 61.6% |
| Kagi LLM Benchmark | — | 76.8% |
| NYT Connections (extended) | 89.6% | — |
| ARC-AGI-1 | — | 70.2% |
| CritPt | 14.3% | — |
| EnigmaEval | — | 18.8% |
| LMArena Hard Prompts | 1483 | — |
| Mystery Game Puzzles | 43% | — |
| DTBench | 89.9% | — |
| LMCA | 47% | — |
| Surface Evolver Bench | 46.3% | — |
Math DeepSeek V4.1 Flash leads
DeepSeek V4.1 Flash: 66.7 (#25), GPT-5 Pro: 48.5 (#63)
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| FrontierMath (Tiers 1-3) | 67.4% | 55.8% |
| FrontierMath Tier 4 | 26.8% | 19.5% |
| OTIS Mock AIME 2024-2025 | 98.3% | — |
| ProofBench | 54% | — |
| LMArena Math | 1477 | — |
| FrontierMath Tier 4 (v1) | — | 14.6% |
Knowledge DeepSeek V4.1 Flash leads
DeepSeek V4.1 Flash: 57.9 (#38), GPT-5 Pro: 56.7 (#42)
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| GPQA Diamond | 89.8% | — |
| Humanity's Last Exam | — | 31.6% |
| LMArena Expert | 1506 | — |
Multimodal Not comparable
DeepSeek V4.1 Flash: 39.1 (#61), GPT-5 Pro: —
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| LMArena Vision | 1277 | — |
| Furniture Assembly | 34.2% | — |
Multilingual Not comparable
DeepSeek V4.1 Flash: 55.0 (#35), GPT-5 Pro: —
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| LMArena Non-English | 1448 | — |
| LMArena Chinese | 1497 | — |
| LMArena French | 1452 | — |
| LMArena German | 1484 | — |
| LMArena Japanese | 1412 | — |
| LMArena Korean | 1452 | — |
| LMArena Russian | 1471 | — |
| LMArena Spanish | 1459 | — |
Instruction Following Not comparable
DeepSeek V4.1 Flash: 77.3 (#26), GPT-5 Pro: —
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| LMArena Instruction Following | 1474 | — |
Long Context Not comparable
DeepSeek V4.1 Flash: 45.2 (#47), GPT-5 Pro: —
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| LMArena Longer Query | 1475 | — |
Writing & Preference Not comparable
DeepSeek V4.1 Flash: 65.4 (#48), GPT-5 Pro: —
| Benchmark | DeepSeek V4.1 Flash | GPT-5 Pro |
|---|---|---|
| LMArena Text | 1462 | — |
| LMArena Creative Writing | 1435 | — |
| EQ-Bench Creative Writing | 1540 | — |
| LMArena Multi-Turn | 1457 | — |
Frequently asked questions
Is DeepSeek V4.1 Flash better than GPT-5 Pro?
DeepSeek V4.1 Flash is the stronger model overall, scoring 52.8 to 46.4 on the Noometry Index.
Which is cheaper, DeepSeek V4.1 Flash or GPT-5 Pro?
DeepSeek V4.1 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; GPT-5 Pro lists at $15 and $120.
Is DeepSeek V4.1 Flash or GPT-5 Pro better for coding?
DeepSeek V4.1 Flash scores higher on coding benchmarks: 52.9 versus 44.0 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4.1 Flash does, with 1M tokens against 400K.
How many benchmarks do DeepSeek V4.1 Flash and GPT-5 Pro share?
3 benchmarks have published results for both models. DeepSeek V4.1 Flash has 37 scored results on Noometry and GPT-5 Pro has 12.