Model comparison
GPT-6 Sol vs Inkling
GPT-6 Sol is the stronger model overall, scoring 61.8 to 44.1 on the Noometry Index. Inkling costs 1.6× less per token, which makes it the better buy when GPT-6 Sol's lead doesn't matter for your workload.
Last verified . 35 shared benchmarks.
Summary
- They share 35 benchmarks with published results for both. GPT-6 Sol scores higher in 6 categories and Inkling in 3 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Sol leads 87.2 to 31.3.
- The biggest single-benchmark swing is FrontierMath Tier 4: 90% for GPT-6 Sol and 4.9% for Inkling.
- Inkling is cheaper at $1.87 / $4.68 per million input/output tokens, against $2 / $10 for GPT-6 Sol.
- GPT-6 Sol accepts more context: 1.05M tokens versus 66K.
- Inkling has downloadable open weights; the other is API-only.
Side by side
| GPT-6 Sol | Inkling | |
|---|---|---|
| Provider | OpenAI | Thinking Machines Lab |
| Noometry Index | 61.8 | 44.1 |
| Released | 2026-09-22 | 2026-07-15 |
| Weights | Proprietary | Open |
| Context window | 1.05M | 66K |
| Max output | 128K | 66K |
| Input $ / M tokens | $2 | $1.87 |
| Output $ / M tokens | $10 | $4.68 |
| Results tracked | 45 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Sol leads
GPT-6 Sol: 60.1 (#11), Inkling: 34.5 (#234)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| FrontierCode | 49.3% | 14% |
| LMArena WebDev | 1688 | 1413 |
| SciCode | 57.6% | 47% |
| LMArena Coding | 1447 | 1464 |
| ALE-Bench | 2,462 | 946 |
| DeepSWE | 68.8% | — |
| FrontierSWE | — | 4.1% |
| WeirdML | — | 32.3% |
Agentic & Tool Use GPT-6 Sol leads
GPT-6 Sol: 37.2 (#36), Inkling: 29.6 (#85)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| APEX-Agents | 54.3% | 33.8% |
| τ²-bench Banking | — | 25% |
| GDP.pdf | 26.4% | — |
| Vending-Bench 2 | 14,428 | — |
Reasoning GPT-6 Sol leads
GPT-6 Sol: 74.0 (#9), Inkling: 40.4 (#56)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| ARC-AGI-2 | 89.6% | 36.5% |
| ARC-AGI-1 | 95.5% | 79.5% |
| CritPt | 30.9% | 5.4% |
| LMArena Hard Prompts | 1418 | 1451 |
| DTBench | 97.3% | 87.5% |
| LMCA | 59.1% | 37.6% |
| Epoch Capabilities Index | 162.72 | 148.54 |
| SimpleBench | — | 50% |
| NYT Connections (extended) | 90.1% | — |
| Chess Puzzles | — | 21% |
| EBR-Bench | 53.3% | — |
| Mystery Game Puzzles | 56% | — |
Math GPT-6 Sol leads
GPT-6 Sol: 87.2 (#7), Inkling: 31.3 (#225)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| FrontierMath (Tiers 1-3) | 89.8% | 33.3% |
| FrontierMath Tier 4 | 90% | 4.9% |
| OTIS Mock AIME 2024-2025 | 100% | 88.9% |
| ProofBench | 83% | 0% |
| LMArena Math | 1402 | 1479 |
Knowledge GPT-6 Sol leads
GPT-6 Sol: 64.8 (#15), Inkling: 55.1 (#49)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| GPQA Diamond | 94.3% | 88.3% |
| SimpleQA Verified | 60.7% | 40.3% |
| LMArena Expert | 1439 | 1465 |
| Vectara Hallucination Rate | 6.5% | — |
Multimodal Not comparable
GPT-6 Sol: 47.6 (#10), Inkling: —
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| LMArena Vision | 1245 | — |
| Blueprint-Bench 2 | 36.9% | — |
| Furniture Assembly | 58.3% | — |
Multilingual Inkling leads
GPT-6 Sol: 50.5 (#118), Inkling: 54.0 (#52)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| LMArena Non-English | 1385 | 1434 |
| LMArena Chinese | 1405 | 1490 |
| LMArena French | 1410 | 1458 |
| LMArena German | 1390 | 1446 |
| LMArena Japanese | 1385 | 1429 |
| LMArena Korean | 1341 | 1404 |
| LMArena Russian | 1401 | 1429 |
| LMArena Spanish | 1384 | 1448 |
Instruction Following Too close to call
GPT-6 Sol: 74.5 (#94), Inkling: 75.1 (#71)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| LMArena Instruction Following | 1412 | 1426 |
Long Context Too close to call
GPT-6 Sol: 43.1 (#108), Inkling: 43.8 (#86)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| LMArena Longer Query | 1411 | 1434 |
Writing & Preference GPT-6 Sol leads
GPT-6 Sol: 71.9 (#18), Inkling: 65.2 (#51)
| Benchmark | GPT-6 Sol | Inkling |
|---|---|---|
| LMArena Text | 1395 | 1441 |
| LMArena Creative Writing | 1378 | 1387 |
| EQ-Bench Creative Writing | 2125 | 1611 |
| LMArena Multi-Turn | 1412 | 1436 |
| EQ-Bench 4 | — | 1226 |
Frequently asked questions
Is GPT-6 Sol better than Inkling?
GPT-6 Sol is the stronger model overall, scoring 61.8 to 44.1 on the Noometry Index. Inkling costs 1.6× less per token, which makes it the better buy when GPT-6 Sol's lead doesn't matter for your workload.
Which is cheaper, GPT-6 Sol or Inkling?
Inkling is cheaper. It lists at $1.87 per million input tokens and $4.68 per million output tokens; GPT-6 Sol lists at $2 and $10.
Is GPT-6 Sol or Inkling better for coding?
GPT-6 Sol scores higher on coding benchmarks: 60.1 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Sol does, with 1.05M tokens against 66K.
How many benchmarks do GPT-6 Sol and Inkling share?
35 benchmarks have published results for both models. GPT-6 Sol has 45 scored results on Noometry and Inkling has 41.