Model comparison
gpt-oss-20b vs Inkling
Inkling is the stronger model overall, scoring 44.1 to 32.5 on the Noometry Index. gpt-oss-20b costs 71× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Last verified . 27 shared benchmarks.
Summary
- They share 27 benchmarks with published results for both. gpt-oss-20b scores higher in 2 categories and Inkling in 7 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Inkling leads 65.2 to 35.5.
- The biggest single-benchmark swing is GPQA Diamond: 60.8% for gpt-oss-20b and 88.3% for Inkling.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- gpt-oss-20b accepts more context: 131K tokens versus 66K.
Side by side
| gpt-oss-20b | Inkling | |
|---|---|---|
| Provider | OpenAI | Thinking Machines Lab |
| Noometry Index | 32.5 | 44.1 |
| Released | 2025-08-05 | 2026-07-15 |
| Weights | Open | Open |
| Context window | 131K | 66K |
| Max output | 16K | 66K |
| Input $ / M tokens | $0.018 | $1.87 |
| Output $ / M tokens | $0.09 | $4.68 |
| Results tracked | 34 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding gpt-oss-20b leads
gpt-oss-20b: 37.6 (#192), Inkling: 34.5 (#234)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| SciCode | 34.4% | 47% |
| WeirdML | 40.9% | 32.3% |
| LMArena Coding | 1306 | 1464 |
| ALE-Bench | 566.05 | 946 |
| FrontierCode | — | 14% |
| LMArena WebDev | — | 1413 |
| FrontierSWE | — | 4.1% |
Agentic & Tool Use Inkling leads
gpt-oss-20b: 9.3 (#154), Inkling: 29.6 (#85)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| Terminal-Bench | 3.4% | — |
| APEX-Agents | — | 33.8% |
| τ²-bench Banking | — | 25% |
Reasoning Inkling leads
gpt-oss-20b: 19.3 (#261), Inkling: 40.4 (#56)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| CritPt | 1.4% | 5.4% |
| Chess Puzzles | 4% | 21% |
| LMArena Hard Prompts | 1274 | 1451 |
| DTBench | 68% | 87.5% |
| LMCA | 14.5% | 37.6% |
| Epoch Capabilities Index | 137.82 | 148.54 |
| ARC-AGI-2 | — | 36.5% |
| SimpleBench | — | 50% |
| Kagi LLM Benchmark | 53.2% | — |
| ARC-AGI-1 | — | 79.5% |
Math gpt-oss-20b leads
gpt-oss-20b: 39.4 (#103), Inkling: 31.3 (#225)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 65.3% | 88.9% |
| LMArena Math | 1317 | 1479 |
| FrontierMath (Tiers 1-3) | — | 33.3% |
| FrontierMath Tier 4 | — | 4.9% |
| ProofBench | — | 0% |
| Omni-MATH | 56.5% | — |
Knowledge Inkling leads
gpt-oss-20b: 34.6 (#195), Inkling: 55.1 (#49)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| GPQA Diamond | 60.8% | 88.3% |
| LMArena Expert | 1258 | 1465 |
| SimpleQA Verified | — | 40.3% |
| MMLU-Pro | 74% | — |
| GPQA (HELM) | 59.4% | — |
Multilingual Inkling leads
gpt-oss-20b: 42.2 (#197), Inkling: 54.0 (#52)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| LMArena Non-English | 1268 | 1434 |
| LMArena Chinese | 1314 | 1490 |
| LMArena German | 1255 | 1446 |
| LMArena Japanese | 1244 | 1429 |
| LMArena Korean | 1236 | 1404 |
| LMArena Russian | 1278 | 1429 |
| LMArena Spanish | 1267 | 1448 |
| LMArena French | — | 1458 |
Instruction Following Inkling leads
gpt-oss-20b: 61.8 (#240), Inkling: 75.1 (#71)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| LMArena Instruction Following | 1236 | 1426 |
| IFEval | 73.2% | — |
Long Context Inkling leads
gpt-oss-20b: 37.9 (#209), Inkling: 43.8 (#86)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| LMArena Longer Query | 1250 | 1434 |
Writing & Preference Inkling leads
gpt-oss-20b: 35.5 (#265), Inkling: 65.2 (#51)
| Benchmark | gpt-oss-20b | Inkling |
|---|---|---|
| LMArena Text | 1287 | 1441 |
| LMArena Creative Writing | 1201 | 1387 |
| EQ-Bench Creative Writing | 666 | 1611 |
| LMArena Multi-Turn | 1268 | 1436 |
| WildBench | 73.7% | — |
| EQ-Bench 4 | — | 1226 |
Frequently asked questions
Is gpt-oss-20b better than Inkling?
Inkling is the stronger model overall, scoring 44.1 to 32.5 on the Noometry Index. gpt-oss-20b costs 71× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Which is cheaper, gpt-oss-20b or Inkling?
gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; Inkling lists at $1.87 and $4.68.
Is gpt-oss-20b or Inkling better for coding?
gpt-oss-20b scores higher on coding benchmarks: 37.6 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
gpt-oss-20b does, with 131K tokens against 66K.
How many benchmarks do gpt-oss-20b and Inkling share?
27 benchmarks have published results for both models. gpt-oss-20b has 34 scored results on Noometry and Inkling has 41.