# DeepSeek V4 Pro vs Inkling

> DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 44.1 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-pro-vs-inkling
- Last updated: 2026-10-10
- Shared benchmarks: 38

## Summary

- They share 38 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 9 categories and Inkling in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Pro leads 64.8 to 31.3.
- The biggest single-benchmark swing is ProofBench: 50% for DeepSeek V4 Pro and 0% for Inkling.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 66K.

## Snapshot

| | DeepSeek V4 Pro | Inkling |
|---|---|---|
| Provider | DeepSeek | Thinking Machines Lab |
| Noometry Index | 54.3 | 44.1 |
| Rank | 31 | 80 |
| Context | 1M | 66K |
| Input $/M | $0.66 | $1.87 |
| Output $/M | $1.98 | $4.68 |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Pro: 52.4 (#34)
- Inkling: 34.5 (#234)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| FrontierCode | 28.6% | 14% |
| LMArena WebDev | 1582 | 1413 |
| SciCode | 51% | 47% |
| WeirdML | 66.2% | 32.3% |
| LMArena Coding | 1470 | 1464 |
| ALE-Bench | 1,403 | 946 |
| SWE-bench Verified | 77.6% | — |
| FrontierSWE | — | 4.1% |

## Agentic & Tool Use

- DeepSeek V4 Pro: 32.8 (#58)
- Inkling: 29.6 (#85)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| APEX-Agents | 47.3% | 33.8% |
| τ²-bench Banking | — | 25% |
| Vending-Bench 2 | 3,285 | — |

## Reasoning

- DeepSeek V4 Pro: 56.5 (#24)
- Inkling: 40.4 (#56)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| ARC-AGI-2 | 61.3% | 36.5% |
| ARC-AGI-1 | 90.5% | 79.5% |
| CritPt | 18% | 5.4% |
| Chess Puzzles | 47% | 21% |
| LMArena Hard Prompts | 1461 | 1451 |
| DTBench | 93.9% | 87.5% |
| LMCA | 45.5% | 37.6% |
| Epoch Capabilities Index | 155.31 | 148.54 |
| SimpleBench | — | 50% |
| Kagi LLM Benchmark | 53.5% | — |
| NYT Connections (extended) | 91.3% | — |
| Mystery Game Puzzles | 43% | — |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |

## Math

- DeepSeek V4 Pro: 64.8 (#30)
- Inkling: 31.3 (#225)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 33.3% |
| FrontierMath Tier 4 | 26.8% | 4.9% |
| OTIS Mock AIME 2024-2025 | 98.6% | 88.9% |
| ProofBench | 50% | 0% |
| LMArena Math | 1455 | 1479 |
| MathArena Final-Answer Competitions | 76.6% | — |

## Knowledge

- DeepSeek V4 Pro: 59.5 (#31)
- Inkling: 55.1 (#49)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| GPQA Diamond | 91.7% | 88.3% |
| SimpleQA Verified | 52.9% | 40.3% |
| LMArena Expert | 1464 | 1465 |
| Vectara Hallucination Rate | 8.6% | — |

## Multilingual

- DeepSeek V4 Pro: 54.4 (#45)
- Inkling: 54.0 (#52)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| LMArena Non-English | 1439 | 1434 |
| LMArena Chinese | 1486 | 1490 |
| LMArena French | 1472 | 1458 |
| LMArena German | 1458 | 1446 |
| LMArena Japanese | 1445 | 1429 |
| LMArena Korean | 1447 | 1404 |
| LMArena Russian | 1453 | 1429 |
| LMArena Spanish | 1458 | 1448 |

## Instruction Following

- DeepSeek V4 Pro: 76.1 (#47)
- Inkling: 75.1 (#71)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| LMArena Instruction Following | 1448 | 1426 |

## Long Context

- DeepSeek V4 Pro: 45.0 (#51)
- Inkling: 43.8 (#86)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| LMArena Longer Query | 1458 | 1434 |
| CL-bench Life | 13.5% | — |

## Writing & Preference

- DeepSeek V4 Pro: 65.5 (#46)
- Inkling: 65.2 (#51)

| Benchmark | DeepSeek V4 Pro | Inkling |
|---|---|---|
| LMArena Text | 1451 | 1441 |
| LMArena Creative Writing | 1446 | 1387 |
| EQ-Bench Creative Writing | 1553 | 1611 |
| EQ-Bench 4 | 1166 | 1226 |
| LMArena Multi-Turn | 1467 | 1436 |

## FAQ

### Is DeepSeek V4 Pro better than Inkling?

DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 44.1 on the Noometry Index.

### Which is cheaper, DeepSeek V4 Pro or Inkling?

DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; Inkling lists at $1.87 and $4.68.

### Is DeepSeek V4 Pro or Inkling better for coding?

DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 34.5 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Pro does, with 1M tokens against 66K.

### How many benchmarks do DeepSeek V4 Pro and Inkling share?

38 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Inkling has 41.