Model comparison
DeepSeek V4 Flash vs Grok 4.5
Grok 4.5 is the stronger model overall, scoring 55.0 to 53.6 on the Noometry Index. DeepSeek V4 Flash costs 11× less per token, which makes it the better buy when Grok 4.5's lead doesn't matter for your workload.
Last verified . 39 shared benchmarks.
DeepSeekSummary
- They share 39 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 0 categories and Grok 4.5 in 8 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Grok 4.5 leads 62.3 to 55.4.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 52.2% for DeepSeek V4 Flash and 83.5% for Grok 4.5.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $2 / $6 for Grok 4.5.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 500K.
- DeepSeek V4 Flash has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4 Flash | Grok 4.5 | |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 53.6 | 55.0 |
| Released | 2026-04-24 | 2026-07-08 |
| Weights | Open | Proprietary |
| Context window | 1M | 500K |
| Max output | 393K | 500K |
| Input $ / M tokens | $0.15 | $2 |
| Output $ / M tokens | $0.60 | $6 |
| Results tracked | 41 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.5 leads
DeepSeek V4 Flash: 47.9 (#59), Grok 4.5: 52.2 (#35)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| FrontierCode | 18.8% | 42.4% |
| LMArena WebDev | 1582 | 1553 |
| SciCode | 49.9% | 54.1% |
| WeirdML | 63% | 46.4% |
| LMArena Coding | 1457 | 1474 |
| ALE-Bench | 1,306 | 1,309 |
| DeepSWE | — | 53.8% |
Agentic & Tool Use Not comparable
DeepSeek V4 Flash: —, Grok 4.5: 44.4 (#17)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| APEX-Agents | — | 56.2% |
| τ²-bench Banking | — | 47.9% |
| PostTrainBench | — | 23.4% |
| GBAEval | — | 65.4% |
| GDP.pdf | — | 14% |
| LMArena Search | — | 1213 |
| Vending-Bench 2 | — | 3,887 |
Reasoning Grok 4.5 leads
DeepSeek V4 Flash: 53.7 (#30), Grok 4.5: 56.1 (#25)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| ARC-AGI-2 | 61.4% | 52.6% |
| SimpleBench | 61.1% | 70% |
| Kagi LLM Benchmark | 52.2% | 83.5% |
| NYT Connections (extended) | 89.6% | 79.9% |
| ARC-AGI-1 | 89% | 87.2% |
| CritPt | 16.6% | 15.4% |
| Chess Puzzles | 33% | 36% |
| LMArena Hard Prompts | 1444 | 1462 |
| DTBench | 90.9% | 96.5% |
| LMCA | 41.7% | 45.2% |
| Epoch Capabilities Index | 154.49 | 153.92 |
| Mystery Game Puzzles | 34% | — |
| Surface Evolver Bench | — | 74.4% |
Math Too close to call
DeepSeek V4 Flash: 60.3 (#37), Grok 4.5: 60.9 (#35)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 57.5% | 57.2% |
| FrontierMath Tier 4 | 24.4% | 24.4% |
| OTIS Mock AIME 2024-2025 | 94.4% | 97.8% |
| ProofBench | 56% | 31% |
| LMArena Math | 1427 | 1459 |
| MathArena Final-Answer Competitions | 76.5% | — |
Knowledge Grok 4.5 leads
DeepSeek V4 Flash: 55.4 (#48), Grok 4.5: 62.3 (#24)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| GPQA Diamond | 91% | 93.4% |
| SimpleQA Verified | 33.6% | 48.3% |
| LMArena Expert | 1441 | 1466 |
Multimodal Not comparable
DeepSeek V4 Flash: —, Grok 4.5: 37.6 (#72)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| LMArena Vision | — | 1288 |
| Blueprint-Bench 2 | — | 27.3% |
| Furniture Assembly | — | 22.5% |
| LMArena Document | — | 1452 |
Multilingual Grok 4.5 leads
DeepSeek V4 Flash: 53.0 (#72), Grok 4.5: 54.4 (#42)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| LMArena Non-English | 1420 | 1440 |
| LMArena Chinese | 1468 | 1496 |
| LMArena French | 1439 | 1456 |
| LMArena German | 1418 | 1446 |
| LMArena Japanese | 1406 | 1428 |
| LMArena Korean | 1384 | 1404 |
| LMArena Russian | 1428 | 1448 |
| LMArena Spanish | 1436 | 1450 |
Instruction Following Grok 4.5 leads
DeepSeek V4 Flash: 74.9 (#81), Grok 4.5: 76.0 (#48)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| LMArena Instruction Following | 1421 | 1446 |
Long Context Too close to call
DeepSeek V4 Flash: 43.8 (#85), Grok 4.5: 44.8 (#56)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| LMArena Longer Query | 1434 | 1463 |
Writing & Preference Grok 4.5 leads
DeepSeek V4 Flash: 63.8 (#61), Grok 4.5: 65.8 (#42)
| Benchmark | DeepSeek V4 Flash | Grok 4.5 |
|---|---|---|
| LMArena Text | 1432 | 1448 |
| LMArena Creative Writing | 1403 | 1442 |
| EQ-Bench Creative Writing | 1559 | 1579 |
| LMArena Multi-Turn | 1449 | 1456 |
Frequently asked questions
Is DeepSeek V4 Flash better than Grok 4.5?
Grok 4.5 is the stronger model overall, scoring 55.0 to 53.6 on the Noometry Index. DeepSeek V4 Flash costs 11× less per token, which makes it the better buy when Grok 4.5's lead doesn't matter for your workload.
Which is cheaper, DeepSeek V4 Flash or Grok 4.5?
DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Grok 4.5 lists at $2 and $6.
Is DeepSeek V4 Flash or Grok 4.5 better for coding?
Grok 4.5 scores higher on coding benchmarks: 52.2 versus 47.9 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4 Flash does, with 1M tokens against 500K.
How many benchmarks do DeepSeek V4 Flash and Grok 4.5 share?
39 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Grok 4.5 has 52.