Model comparison
Claude Sonnet 4.6 vs Inkling
Claude Sonnet 4.6 is the stronger model overall, scoring 50.3 to 44.1 on the Noometry Index. Inkling costs 2.3× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.
Last verified . 36 shared benchmarks.
Summary
- They share 36 benchmarks with published results for both. Claude Sonnet 4.6 scores higher in 8 categories and Inkling in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Sonnet 4.6 leads 52.9 to 31.3.
- The biggest single-benchmark swing is ProofBench: 45% for Claude Sonnet 4.6 and 0% for Inkling.
- Inkling is cheaper at $1.87 / $4.68 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.6.
- Claude Sonnet 4.6 accepts more context: 1M tokens versus 66K.
- Inkling has downloadable open weights; the other is API-only.
Side by side
| Claude Sonnet 4.6 | Inkling | |
|---|---|---|
| Provider | Anthropic | Thinking Machines Lab |
| Noometry Index | 50.3 | 44.1 |
| Released | 2026-02-17 | 2026-07-15 |
| Weights | Proprietary | Open |
| Context window | 1M | 66K |
| Max output | 128K | 66K |
| Input $ / M tokens | $3 | $1.87 |
| Output $ / M tokens | $15 | $4.68 |
| Results tracked | 57 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 46.3 (#67), Inkling: 34.5 (#234)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| FrontierCode | 24.3% | 14% |
| LMArena WebDev | 1522 | 1413 |
| SciCode | 46.8% | 47% |
| WeirdML | 66.1% | 32.3% |
| LMArena Coding | 1504 | 1464 |
| ALE-Bench | 1,327 | 946 |
| SWE-bench Verified | 75.2% | — |
| DeepSWE | 29.9% | — |
| FrontierSWE | — | 4.1% |
Agentic & Tool Use Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 39.1 (#28), Inkling: 29.6 (#85)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| APEX-Agents | 43% | 33.8% |
| Terminal-Bench | 53.4% | — |
| OSWorld 2.0 | 9.3% | — |
| τ²-bench Banking | — | 25% |
| DeepResearch Bench | 54.9% | — |
| OSWorld | 72.1% | — |
| ExploitBench | 23.6% | — |
| GBAEval | 48.8% | — |
| GDP.pdf | 18% | — |
| LMArena Search | 1221 | — |
| Vending-Bench 2 | 7,204 | — |
Reasoning Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 46.1 (#45), Inkling: 40.4 (#56)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| ARC-AGI-2 | 60.4% | 36.5% |
| ARC-AGI-1 | 86.5% | 79.5% |
| CritPt | 3.1% | 5.4% |
| Chess Puzzles | 13% | 21% |
| LMArena Hard Prompts | 1484 | 1451 |
| DTBench | 89.9% | 87.5% |
| LMCA | 46.5% | 37.6% |
| Epoch Capabilities Index | 152.24 | 148.54 |
| SimpleBench | — | 50% |
| NYT Connections (extended) | 80.9% | — |
| Thematic Generalization | 76.3% | — |
| Mystery Game Puzzles | 16% | — |
| ForecastBench | 62 | — |
Math Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 52.9 (#49), Inkling: 31.3 (#225)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 85.8% | 88.9% |
| ProofBench | 45% | 0% |
| LMArena Math | 1462 | 1479 |
| FrontierMath (Tiers 1-3) | — | 33.3% |
| FrontierMath Tier 4 | — | 4.9% |
| FrontierMath (Feb 2025 set) | 32.4% | — |
| FrontierMath Tier 4 (v1) | 8.3% | — |
Knowledge Inkling leads
Claude Sonnet 4.6: 51.7 (#65), Inkling: 55.1 (#49)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| GPQA Diamond | 87.4% | 88.3% |
| SimpleQA Verified | 35.5% | 40.3% |
| LMArena Expert | 1500 | 1465 |
| Vectara Hallucination Rate | 10.6% | — |
Multimodal Not comparable
Claude Sonnet 4.6: 38.0 (#68), Inkling: —
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| LMArena Vision | 1283 | — |
| Blueprint-Bench 2 | 6.7% | — |
| LMArena Document | 1482 | — |
Multilingual Too close to call
Claude Sonnet 4.6: 54.4 (#41), Inkling: 54.0 (#52)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| LMArena Non-English | 1440 | 1434 |
| LMArena Chinese | 1491 | 1490 |
| LMArena French | 1465 | 1458 |
| LMArena German | 1428 | 1446 |
| LMArena Japanese | 1420 | 1429 |
| LMArena Korean | 1411 | 1404 |
| LMArena Russian | 1440 | 1429 |
| LMArena Spanish | 1464 | 1448 |
Instruction Following Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 77.4 (#25), Inkling: 75.1 (#71)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| LMArena Instruction Following | 1475 | 1426 |
Long Context Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 45.3 (#44), Inkling: 43.8 (#86)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| LMArena Longer Query | 1479 | 1434 |
Writing & Preference Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 70.2 (#22), Inkling: 65.2 (#51)
| Benchmark | Claude Sonnet 4.6 | Inkling |
|---|---|---|
| LMArena Text | 1458 | 1441 |
| LMArena Creative Writing | 1435 | 1387 |
| EQ-Bench Creative Writing | 1810 | 1611 |
| EQ-Bench 4 | 1207 | 1226 |
| LMArena Multi-Turn | 1464 | 1436 |
Frequently asked questions
Is Claude Sonnet 4.6 better than Inkling?
Claude Sonnet 4.6 is the stronger model overall, scoring 50.3 to 44.1 on the Noometry Index. Inkling costs 2.3× less per token, which makes it the better buy when Claude Sonnet 4.6's lead doesn't matter for your workload.
Which is cheaper, Claude Sonnet 4.6 or Inkling?
Inkling is cheaper. It lists at $1.87 per million input tokens and $4.68 per million output tokens; Claude Sonnet 4.6 lists at $3 and $15.
Is Claude Sonnet 4.6 or Inkling better for coding?
Claude Sonnet 4.6 scores higher on coding benchmarks: 46.3 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
Claude Sonnet 4.6 does, with 1M tokens against 66K.
How many benchmarks do Claude Sonnet 4.6 and Inkling share?
36 benchmarks have published results for both models. Claude Sonnet 4.6 has 57 scored results on Noometry and Inkling has 41.