Model comparison
DeepSeek V4 Flash vs GPT-5.4 Pro
GPT-5.4 Pro is the stronger model overall, scoring 58.9 to 53.6 on the Noometry Index. DeepSeek V4 Flash costs 257× less per token, which makes it the better buy when GPT-5.4 Pro's lead doesn't matter for your workload.
Last verified . 11 shared benchmarks.
DeepSeekSummary
- They share 11 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 1 category and GPT-5.4 Pro in 3 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.4 Pro leads 70.7 to 53.7.
- The biggest single-benchmark swing is FrontierMath Tier 4: 24.4% for DeepSeek V4 Flash and 58.5% for GPT-5.4 Pro.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $30 / $180 for GPT-5.4 Pro.
- GPT-5.4 Pro accepts more context: 1.05M tokens versus 1M.
- DeepSeek V4 Flash has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4 Flash | GPT-5.4 Pro | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 53.6 | 58.9 |
| Released | 2026-04-24 | 2026-03-05 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 393K | 128K |
| Input $ / M tokens | $0.15 | $30 |
| Output $ / M tokens | $0.60 | $180 |
| Results tracked | 41 | 16 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Flash leads
DeepSeek V4 Flash: 47.9 (#59), GPT-5.4 Pro: 43.2 (#90)
| Benchmark | DeepSeek V4 Flash | GPT-5.4 Pro |
|---|---|---|
| WeirdML | 63% | 57.4% |
| FrontierCode | 18.8% | — |
| LMArena WebDev | 1582 | — |
| SciCode | 49.9% | — |
| LMArena Coding | 1457 | — |
| ALE-Bench | 1,306 | — |
Reasoning GPT-5.4 Pro leads
DeepSeek V4 Flash: 53.7 (#30), GPT-5.4 Pro: 70.7 (#13)
| Benchmark | DeepSeek V4 Flash | GPT-5.4 Pro |
|---|---|---|
| ARC-AGI-2 | 61.4% | 83.3% |
| SimpleBench | 61.1% | 74.1% |
| ARC-AGI-1 | 89% | 94.5% |
| CritPt | 16.6% | 30% |
| Chess Puzzles | 33% | 58.6% |
| Epoch Capabilities Index | 154.49 | 158.93 |
| Kagi LLM Benchmark | 52.2% | — |
| NYT Connections (extended) | 89.6% | — |
| EnigmaEval | — | 23.8% |
| LMArena Hard Prompts | 1444 | — |
| Mystery Game Puzzles | 34% | — |
| DTBench | 90.9% | — |
| LMCA | 41.7% | — |
Math GPT-5.4 Pro leads
DeepSeek V4 Flash: 60.3 (#37), GPT-5.4 Pro: 72.4 (#22)
| Benchmark | DeepSeek V4 Flash | GPT-5.4 Pro |
|---|---|---|
| FrontierMath (Tiers 1-3) | 57.5% | 82.5% |
| FrontierMath Tier 4 | 24.4% | 58.5% |
| MathArena Final-Answer Competitions | 76.5% | — |
| OTIS Mock AIME 2024-2025 | 94.4% | — |
| ProofBench | 56% | — |
| LMArena Math | 1427 | — |
| FrontierMath (Feb 2025 set) | — | 50% |
| FrontierMath Tier 4 (v1) | — | 37.5% |
Knowledge GPT-5.4 Pro leads
DeepSeek V4 Flash: 55.4 (#48), GPT-5.4 Pro: 68.3 (#7)
| Benchmark | DeepSeek V4 Flash | GPT-5.4 Pro |
|---|---|---|
| GPQA Diamond | 91% | 94.6% |
| SimpleQA Verified | 33.6% | 46.3% |
| Humanity's Last Exam | — | 44.3% |
| Vectara Hallucination Rate | — | 8.3% |
| LMArena Expert | 1441 | — |
Multilingual Not comparable
DeepSeek V4 Flash: 53.0 (#72), GPT-5.4 Pro: —
| Benchmark | DeepSeek V4 Flash | GPT-5.4 Pro |
|---|---|---|
| LMArena Non-English | 1420 | — |
| LMArena Chinese | 1468 | — |
| LMArena French | 1439 | — |
| LMArena German | 1418 | — |
| LMArena Japanese | 1406 | — |
| LMArena Korean | 1384 | — |
| LMArena Russian | 1428 | — |
| LMArena Spanish | 1436 | — |
Instruction Following Not comparable
DeepSeek V4 Flash: 74.9 (#81), GPT-5.4 Pro: —
| Benchmark | DeepSeek V4 Flash | GPT-5.4 Pro |
|---|---|---|
| LMArena Instruction Following | 1421 | — |
Long Context Not comparable
DeepSeek V4 Flash: 43.8 (#85), GPT-5.4 Pro: —
| Benchmark | DeepSeek V4 Flash | GPT-5.4 Pro |
|---|---|---|
| LMArena Longer Query | 1434 | — |
Writing & Preference Not comparable
DeepSeek V4 Flash: 63.8 (#61), GPT-5.4 Pro: —
| Benchmark | DeepSeek V4 Flash | GPT-5.4 Pro |
|---|---|---|
| LMArena Text | 1432 | — |
| LMArena Creative Writing | 1403 | — |
| EQ-Bench Creative Writing | 1559 | — |
| LMArena Multi-Turn | 1449 | — |
Frequently asked questions
Is DeepSeek V4 Flash better than GPT-5.4 Pro?
GPT-5.4 Pro is the stronger model overall, scoring 58.9 to 53.6 on the Noometry Index. DeepSeek V4 Flash costs 257× less per token, which makes it the better buy when GPT-5.4 Pro's lead doesn't matter for your workload.
Which is cheaper, DeepSeek V4 Flash or GPT-5.4 Pro?
DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; GPT-5.4 Pro lists at $30 and $180.
Is DeepSeek V4 Flash or GPT-5.4 Pro better for coding?
DeepSeek V4 Flash scores higher on coding benchmarks: 47.9 versus 43.2 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 Pro does, with 1.05M tokens against 1M.
How many benchmarks do DeepSeek V4 Flash and GPT-5.4 Pro share?
11 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and GPT-5.4 Pro has 16.