# DeepSeek V4 Flash vs MiniMax-M3

> DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 43.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-flash-vs-minimax-m3
- Last updated: 2026-10-10
- Shared benchmarks: 32

## Summary

- They share 32 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 4 categories and MiniMax-M3 in 4 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Flash leads 53.7 to 29.3.
- The biggest single-benchmark swing is ProofBench: 56% for DeepSeek V4 Flash and 18% for MiniMax-M3.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.30 / $1.20 for MiniMax-M3.

## Snapshot

| | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| Provider | DeepSeek | MiniMax |
| Noometry Index | 53.6 | 43.5 |
| Rank | 32 | 81 |
| Context | 1M | 1M |
| Input $/M | $0.15 | $0.30 |
| Output $/M | $0.60 | $1.20 |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Flash: 47.6 (#56)
- MiniMax-M3: 41.5 (#111)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| FrontierCode | 18.8% | 14.7% |
| LMArena WebDev | 1581 | 1482 |
| SciCode | 49.9% | 47.1% |
| LMArena Coding | 1457 | 1470 |
| ALE-Bench | 1,306 | 640.02 |
| WeirdML | 63% | — |

## Agentic & Tool Use

- DeepSeek V4 Flash: —
- MiniMax-M3: 22.5 (#121)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| APEX-Agents | — | 37.7% |
| OSWorld 2.0 | — | 4.6% |
| GBAEval | — | 0.9% |
| Vending-Bench 2 | — | 2,158 |

## Reasoning

- DeepSeek V4 Flash: 53.7 (#25)
- MiniMax-M3: 29.3 (#83)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| SimpleBench | 61.1% | 45.8% |
| NYT Connections (extended) | 89.6% | 65.1% |
| CritPt | 16.6% | 3.7% |
| Chess Puzzles | 33% | 14% |
| LMArena Hard Prompts | 1443 | 1447 |
| Mystery Game Puzzles | 34% | 8% |
| DTBench | 90.9% | 78.9% |
| LMCA | 41.7% | 33.7% |
| Epoch Capabilities Index | 154.51 | 147 |
| ARC-AGI-2 | 61.4% | — |
| Kagi LLM Benchmark | 52.2% | — |
| ARC-AGI-1 | 89% | — |
| Surface Evolver Bench | — | 55% |
| ForecastBench | — | 61.4 |

## Math

- DeepSeek V4 Flash: 60.0 (#31)
- MiniMax-M3: 39.1 (#83)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 94.4% | 71.1% |
| ProofBench | 56% | 18% |
| LMArena Math | 1427 | 1432 |
| FrontierMath (Tiers 1-3) | 57.5% | — |
| FrontierMath Tier 4 | 24.4% | — |
| MathArena Final-Answer Competitions | 76.5% | — |

## Knowledge

- DeepSeek V4 Flash: 55.5 (#43)
- MiniMax-M3: 58.4 (#32)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| GPQA Diamond | 91% | 90.9% |
| LMArena Expert | 1441 | 1462 |
| SimpleQA Verified | 33.6% | — |

## Multimodal

- DeepSeek V4 Flash: —
- MiniMax-M3: 39.5 (#45)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| LMArena Vision | — | 1254 |
| LMArena Document | — | 1435 |

## Multilingual

- DeepSeek V4 Flash: 53.2 (#71)
- MiniMax-M3: 53.2 (#70)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| LMArena Non-English | 1420 | 1420 |
| LMArena Chinese | 1467 | 1465 |
| LMArena French | 1438 | 1445 |
| LMArena German | 1418 | 1424 |
| LMArena Japanese | 1403 | 1381 |
| LMArena Korean | 1383 | 1372 |
| LMArena Russian | 1428 | 1430 |
| LMArena Spanish | 1437 | 1431 |

## Instruction Following

- DeepSeek V4 Flash: 75.1 (#76)
- MiniMax-M3: 75.7 (#56)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| LMArena Instruction Following | 1421 | 1433 |

## Long Context

- DeepSeek V4 Flash: 43.8 (#81)
- MiniMax-M3: 44.2 (#68)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| LMArena Longer Query | 1435 | 1445 |

## Writing & Preference

- DeepSeek V4 Flash: 64.0 (#59)
- MiniMax-M3: 62.3 (#79)

| Benchmark | DeepSeek V4 Flash | MiniMax-M3 |
|---|---|---|
| LMArena Text | 1432 | 1433 |
| LMArena Creative Writing | 1403 | 1405 |
| LMArena Multi-Turn | 1449 | 1441 |
| EQ-Bench Creative Writing | 1559 | — |
| EQ-Bench 4 | — | 1150 |

## FAQ

### Is DeepSeek V4 Flash better than MiniMax-M3?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 43.5 on the Noometry Index.

### Which is cheaper, DeepSeek V4 Flash or MiniMax-M3?

DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; MiniMax-M3 lists at $0.30 and $1.20.

### Is DeepSeek V4 Flash or MiniMax-M3 better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.6 versus 41.5 in the Noometry coding category.

### Which has the bigger context window?

Both accept 1M tokens.

### How many benchmarks do DeepSeek V4 Flash and MiniMax-M3 share?

32 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and MiniMax-M3 has 41.