Model comparison
DeepSeek-R1 vs Inkling
Inkling is the stronger model overall, scoring 44.1 to 42.3 on the Noometry Index. DeepSeek-R1 costs 2.8× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Last verified . 28 shared benchmarks.
DeepSeekSummary
- They share 28 benchmarks with published results for both. DeepSeek-R1 scores higher in 4 categories and Inkling in 5 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Inkling leads 40.4 to 18.6.
- The biggest single-benchmark swing is ARC-AGI-1: 21.2% for DeepSeek-R1 and 79.5% for Inkling.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- DeepSeek-R1 accepts more context: 164K tokens versus 66K.
- Inkling has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1 | Inkling | |
|---|---|---|
| Provider | DeepSeek | Thinking Machines Lab |
| Noometry Index | 42.3 | 44.1 |
| Released | 2025-01-20 | 2026-07-15 |
| Weights | Proprietary | Open |
| Context window | 164K | 66K |
| Max output | 64K | 66K |
| Input $ / M tokens | $0.50 | $1.87 |
| Output $ / M tokens | $2.15 | $4.68 |
| Results tracked | 52 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), Inkling: 34.5 (#234)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| SciCode | 35.7% | 47% |
| WeirdML | 41.6% | 32.3% |
| LMArena Coding | 1427 | 1464 |
| ALE-Bench | 804.12 | 946 |
| FrontierCode | — | 14% |
| Aider Polyglot | 71.4% | — |
| LMArena WebDev | — | 1413 |
| FrontierSWE | — | 4.1% |
| LiveBench Coding | 66.7% | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use DeepSeek-R1 leads
DeepSeek-R1: 30.7 (#75), Inkling: 29.6 (#85)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| APEX-Agents | — | 33.8% |
| τ²-bench Banking | — | 25% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning Inkling leads
DeepSeek-R1: 18.6 (#278), Inkling: 40.4 (#56)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| ARC-AGI-2 | 1.3% | 36.5% |
| SimpleBench | 40.8% | 50% |
| ARC-AGI-1 | 21.2% | 79.5% |
| CritPt | 1.1% | 5.4% |
| LMArena Hard Prompts | 1416 | 1451 |
| Epoch Capabilities Index | 141.29 | 148.54 |
| Kagi LLM Benchmark | 69.4% | — |
| Chess Puzzles | — | 21% |
| LiveBench Reasoning | 83.2% | — |
| DTBench | — | 87.5% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 37.6% |
| ForecastBench | 60 | — |
| LiveBench | 71.6% | — |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), Inkling: 31.3 (#225)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 88.9% |
| LMArena Math | 1400 | 1479 |
| FrontierMath (Tiers 1-3) | — | 33.3% |
| FrontierMath Tier 4 | — | 4.9% |
| ProofBench | — | 0% |
| Omni-MATH | 42.4% | — |
| LiveBench Math | 80.7% | — |
| MATH Level 5 | 96.6% | — |
Knowledge Inkling leads
DeepSeek-R1: 44.5 (#87), Inkling: 55.1 (#49)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| GPQA Diamond | 76.3% | 88.3% |
| LMArena Expert | 1394 | 1465 |
| SimpleQA Verified | — | 40.3% |
| MMLU-Pro | 79.3% | — |
| Confabulations | 12.7% | — |
| Vectara Hallucination Rate | 11.3% | — |
| GPQA (HELM) | 66.6% | — |
Multilingual Inkling leads
DeepSeek-R1: 52.4 (#85), Inkling: 54.0 (#52)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| LMArena Non-English | 1412 | 1434 |
| LMArena Chinese | 1442 | 1490 |
| LMArena French | 1417 | 1458 |
| LMArena German | 1404 | 1446 |
| LMArena Japanese | 1391 | 1429 |
| LMArena Korean | 1360 | 1404 |
| LMArena Russian | 1423 | 1429 |
| LMArena Spanish | 1411 | 1448 |
Instruction Following Inkling leads
DeepSeek-R1: 72.0 (#143), Inkling: 75.1 (#71)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| LMArena Instruction Following | 1382 | 1426 |
| LiveBench Instruction Following | 80.5% | — |
| IFEval | 78.4% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), Inkling: 43.8 (#86)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| LMArena Longer Query | 1391 | 1434 |
| Fiction.LiveBench | 75% | — |
Writing & Preference Inkling leads
DeepSeek-R1: 61.4 (#88), Inkling: 65.2 (#51)
| Benchmark | DeepSeek-R1 | Inkling |
|---|---|---|
| LMArena Text | 1428 | 1441 |
| LMArena Creative Writing | 1405 | 1387 |
| EQ-Bench Creative Writing | 1500 | 1611 |
| LMArena Multi-Turn | 1405 | 1436 |
| Short-Story Creative Writing | 83% | — |
| WildBench | 82.8% | — |
| EQ-Bench 4 | — | 1226 |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than Inkling?
Inkling is the stronger model overall, scoring 44.1 to 42.3 on the Noometry Index. DeepSeek-R1 costs 2.8× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-R1 or Inkling?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; Inkling lists at $1.87 and $4.68.
Is DeepSeek-R1 or Inkling better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-R1 does, with 164K tokens against 66K.
How many benchmarks do DeepSeek-R1 and Inkling share?
28 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Inkling has 41.