Model comparison
DeepSeek-V3.1 vs Inkling
Inkling is the stronger model overall, scoring 44.1 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 6.1× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Last verified . 23 shared benchmarks.
DeepSeekSummary
- They share 23 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 2 categories and Inkling in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Inkling leads 40.4 to 27.9.
- The biggest single-benchmark swing is LMCA: 24.3% for DeepSeek-V3.1 and 37.6% for Inkling.
- DeepSeek-V3.1 is cheaper at $0.25 / $0.95 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- DeepSeek-V3.1 accepts more context: 164K tokens versus 66K.
Side by side
| DeepSeek-V3.1 | Inkling | |
|---|---|---|
| Provider | DeepSeek | Thinking Machines Lab |
| Noometry Index | 42.8 | 44.1 |
| Released | 2025-08-21 | 2026-07-15 |
| Weights | Open | Open |
| Context window | 164K | 66K |
| Max output | 8K | 66K |
| Input $ / M tokens | $0.25 | $1.87 |
| Output $ / M tokens | $0.95 | $4.68 |
| Results tracked | 27 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.1 leads
DeepSeek-V3.1: 40.3 (#144), Inkling: 34.5 (#234)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| WeirdML | 38.4% | 32.3% |
| LMArena Coding | 1417 | 1464 |
| FrontierCode | — | 14% |
| LMArena WebDev | — | 1413 |
| FrontierSWE | — | 4.1% |
| SciCode | — | 47% |
| ALE-Bench | — | 946 |
Agentic & Tool Use Not comparable
DeepSeek-V3.1: —, Inkling: 29.6 (#85)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| APEX-Agents | — | 33.8% |
| τ²-bench Banking | — | 25% |
Reasoning Inkling leads
DeepSeek-V3.1: 27.9 (#110), Inkling: 40.4 (#56)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| SimpleBench | 40% | 50% |
| LMArena Hard Prompts | 1417 | 1451 |
| DTBench | 82.7% | 87.5% |
| LMCA | 24.3% | 37.6% |
| Epoch Capabilities Index | 139.92 | 148.54 |
| ARC-AGI-2 | — | 36.5% |
| Kagi LLM Benchmark | 53.2% | — |
| ARC-AGI-1 | — | 79.5% |
| CritPt | — | 5.4% |
| Chess Puzzles | — | 21% |
| ForecastBench | 58 | — |
Math DeepSeek-V3.1 leads
DeepSeek-V3.1: 38.9 (#122), Inkling: 31.3 (#225)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| LMArena Math | 1420 | 1479 |
| FrontierMath (Tiers 1-3) | — | 33.3% |
| FrontierMath Tier 4 | — | 4.9% |
| OTIS Mock AIME 2024-2025 | — | 88.9% |
| ProofBench | — | 0% |
Knowledge Inkling leads
DeepSeek-V3.1: 43.7 (#90), Inkling: 55.1 (#49)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| LMArena Expert | 1405 | 1465 |
| GPQA Diamond | — | 88.3% |
| SimpleQA Verified | — | 40.3% |
| Vectara Hallucination Rate | 5.5% | — |
Multilingual Inkling leads
DeepSeek-V3.1: 51.6 (#106), Inkling: 54.0 (#52)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| LMArena Non-English | 1400 | 1434 |
| LMArena Chinese | 1469 | 1490 |
| LMArena French | 1447 | 1458 |
| LMArena German | 1411 | 1446 |
| LMArena Japanese | 1378 | 1429 |
| LMArena Korean | 1337 | 1404 |
| LMArena Russian | 1405 | 1429 |
| LMArena Spanish | 1431 | 1448 |
Instruction Following Inkling leads
DeepSeek-V3.1: 73.9 (#110), Inkling: 75.1 (#71)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| LMArena Instruction Following | 1400 | 1426 |
Long Context Inkling leads
DeepSeek-V3.1: 36.3 (#232), Inkling: 43.8 (#86)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| LMArena Longer Query | 1422 | 1434 |
| Fiction.LiveBench | 52.8% | — |
Writing & Preference Inkling leads
DeepSeek-V3.1: 60.3 (#98), Inkling: 65.2 (#51)
| Benchmark | DeepSeek-V3.1 | Inkling |
|---|---|---|
| LMArena Text | 1420 | 1441 |
| LMArena Creative Writing | 1401 | 1387 |
| EQ-Bench Creative Writing | 1436 | 1611 |
| LMArena Multi-Turn | 1408 | 1436 |
| EQ-Bench 4 | — | 1226 |
Frequently asked questions
Is DeepSeek-V3.1 better than Inkling?
Inkling is the stronger model overall, scoring 44.1 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 6.1× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.1 or Inkling?
DeepSeek-V3.1 is cheaper. It lists at $0.25 per million input tokens and $0.95 per million output tokens; Inkling lists at $1.87 and $4.68.
Is DeepSeek-V3.1 or Inkling better for coding?
DeepSeek-V3.1 scores higher on coding benchmarks: 40.3 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3.1 does, with 164K tokens against 66K.
How many benchmarks do DeepSeek-V3.1 and Inkling share?
23 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and Inkling has 41.