Model comparison
DeepSeek V4 Pro vs Inkling-Small
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 46.5 on the Noometry Index. Inkling-Small costs 1.6× less per token, which makes it the better buy when DeepSeek V4 Pro's lead doesn't matter for your workload.
Last verified . 32 shared benchmarks.
DeepSeekSummary
- They share 32 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 8 categories and Inkling-Small in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Pro leads 64.8 to 45.1.
- The biggest single-benchmark swing is ProofBench: 50% for DeepSeek V4 Pro and 6% for Inkling-Small.
- Inkling-Small is cheaper at $0.45 / $1.20 per million input/output tokens, against $0.66 / $1.98 for DeepSeek V4 Pro.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 524K.
Side by side
| DeepSeek V4 Pro | Inkling-Small | |
|---|---|---|
| Provider | DeepSeek | Thinking Machines Lab |
| Noometry Index | 54.3 | 46.5 |
| Released | 2026-04-24 | 2026-07-15 |
| Weights | Open | Open |
| Context window | 1M | 524K |
| Max output | 393K | 1.05M |
| Input $ / M tokens | $0.66 | $0.45 |
| Output $ / M tokens | $1.98 | $1.20 |
| Results tracked | 48 | 33 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Pro leads
DeepSeek V4 Pro: 52.4 (#34), Inkling-Small: 43.6 (#85)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| LMArena WebDev | 1582 | 1409 |
| SciCode | 51% | 48.7% |
| LMArena Coding | 1470 | 1451 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| WeirdML | 66.2% | — |
| ALE-Bench | 1,403 | — |
Agentic & Tool Use Not comparable
DeepSeek V4 Pro: 32.8 (#58), Inkling-Small: —
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| APEX-Agents | 47.3% | — |
| Vending-Bench 2 | 3,285 | — |
Reasoning DeepSeek V4 Pro leads
DeepSeek V4 Pro: 56.5 (#24), Inkling-Small: 38.6 (#63)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| ARC-AGI-2 | 61.3% | 40.1% |
| ARC-AGI-1 | 90.5% | 84% |
| CritPt | 18% | 8.3% |
| Chess Puzzles | 47% | 18% |
| LMArena Hard Prompts | 1461 | 1423 |
| Mystery Game Puzzles | 43% | 6% |
| Epoch Capabilities Index | 155.31 | 150.15 |
| Kagi LLM Benchmark | 53.5% | — |
| NYT Connections (extended) | 91.3% | — |
| DTBench | 93.9% | — |
| LMCA | 45.5% | — |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |
Math DeepSeek V4 Pro leads
DeepSeek V4 Pro: 64.8 (#30), Inkling-Small: 45.1 (#77)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 46.3% |
| FrontierMath Tier 4 | 26.8% | 17.1% |
| OTIS Mock AIME 2024-2025 | 98.6% | 90% |
| ProofBench | 50% | 6% |
| LMArena Math | 1455 | 1459 |
| MathArena Final-Answer Competitions | 76.6% | — |
Knowledge DeepSeek V4 Pro leads
DeepSeek V4 Pro: 59.5 (#31), Inkling-Small: 48.2 (#77)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| GPQA Diamond | 91.7% | 88.5% |
| SimpleQA Verified | 52.9% | 19.1% |
| LMArena Expert | 1464 | 1442 |
| Vectara Hallucination Rate | 8.6% | — |
Multimodal Not comparable
DeepSeek V4 Pro: —, Inkling-Small: 39.1 (#62)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| LMArena Vision | — | 1235 |
Multilingual DeepSeek V4 Pro leads
DeepSeek V4 Pro: 54.4 (#45), Inkling-Small: 51.7 (#104)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| LMArena Non-English | 1439 | 1402 |
| LMArena Chinese | 1486 | 1465 |
| LMArena French | 1472 | 1436 |
| LMArena German | 1458 | 1405 |
| LMArena Japanese | 1445 | 1405 |
| LMArena Korean | 1447 | 1363 |
| LMArena Russian | 1453 | 1391 |
| LMArena Spanish | 1458 | 1428 |
Instruction Following DeepSeek V4 Pro leads
DeepSeek V4 Pro: 76.1 (#47), Inkling-Small: 73.8 (#114)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| LMArena Instruction Following | 1448 | 1399 |
Long Context DeepSeek V4 Pro leads
DeepSeek V4 Pro: 45.0 (#51), Inkling-Small: 42.7 (#118)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| LMArena Longer Query | 1458 | 1401 |
| CL-bench Life | 13.5% | — |
Writing & Preference DeepSeek V4 Pro leads
DeepSeek V4 Pro: 65.5 (#46), Inkling-Small: 59.6 (#107)
| Benchmark | DeepSeek V4 Pro | Inkling-Small |
|---|---|---|
| LMArena Text | 1451 | 1414 |
| LMArena Creative Writing | 1446 | 1331 |
| EQ-Bench Creative Writing | 1553 | 1491 |
| LMArena Multi-Turn | 1467 | 1418 |
| EQ-Bench 4 | 1166 | — |
Frequently asked questions
Is DeepSeek V4 Pro better than Inkling-Small?
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 46.5 on the Noometry Index. Inkling-Small costs 1.6× less per token, which makes it the better buy when DeepSeek V4 Pro's lead doesn't matter for your workload.
Which is cheaper, DeepSeek V4 Pro or Inkling-Small?
Inkling-Small is cheaper. It lists at $0.45 per million input tokens and $1.20 per million output tokens; DeepSeek V4 Pro lists at $0.66 and $1.98.
Is DeepSeek V4 Pro or Inkling-Small better for coding?
DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 43.6 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4 Pro does, with 1M tokens against 524K.
How many benchmarks do DeepSeek V4 Pro and Inkling-Small share?
32 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Inkling-Small has 33.