Model comparison
gpt-oss-20b vs Trinity Large Thinking
Trinity Large Thinking is the stronger model overall, scoring 38.1 to 32.1 on the Noometry Index. gpt-oss-20b costs 11× less per token, which makes it the better buy when Trinity Large Thinking's lead doesn't matter for your workload.
Last verified . 18 shared benchmarks.
Arcee AISummary
- They share 18 benchmarks with published results for both. gpt-oss-20b scores higher in 3 categories and Trinity Large Thinking in 5 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Trinity Large Thinking leads 54.0 to 35.8.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $0.25 / $0.80 for Trinity Large Thinking.
- Trinity Large Thinking accepts more context: 262K tokens versus 131K.
Side by side
| gpt-oss-20b | Trinity Large Thinking | |
|---|---|---|
| Provider | OpenAI | Arcee AI |
| Noometry Index | 32.1 | 38.1 |
| Released | 2025-08-05 | 2026-04-01 |
| Weights | Open | Open |
| Context window | 131K | 262K |
| Max output | 16K | 80K |
| Input $ / M tokens | $0.018 | $0.25 |
| Output $ / M tokens | $0.09 | $0.80 |
| Results tracked | 34 | 24 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding gpt-oss-20b leads
gpt-oss-20b: 37.1 (#185), Trinity Large Thinking: 33.8 (#238)
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| SciCode | 34.4% | 36.1% |
| LMArena Coding | 1307 | 1381 |
| LMArena WebDev | — | 1238 |
| WeirdML | 40.9% | — |
| ALE-Bench | 566.05 | — |
Agentic & Tool Use Not comparable
gpt-oss-20b: 9.2 (#146), Trinity Large Thinking: —
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| Terminal-Bench | 3.4% | — |
Reasoning gpt-oss-20b leads
gpt-oss-20b: 18.8 (#251), Trinity Large Thinking: 16.3 (#298)
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| CritPt | 1.4% | 0.9% |
| LMArena Hard Prompts | 1274 | 1350 |
| Kagi LLM Benchmark | 53.2% | — |
| NYT Connections (extended) | — | 16.5% |
| Chess Puzzles | 4% | — |
| Thematic Generalization | — | 41.6% |
| DTBench | 68% | — |
| LMCA | 14.5% | — |
| Surface Evolver Bench | — | 15.6% |
| Epoch Capabilities Index | 137.81 | — |
Math gpt-oss-20b leads
gpt-oss-20b: 37.6 (#93), Trinity Large Thinking: 34.9 (#149)
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| LMArena Math | 1317 | 1366 |
| OTIS Mock AIME 2024-2025 | 65.3% | — |
| Omni-MATH | 56.5% | — |
Knowledge Trinity Large Thinking leads
gpt-oss-20b: 34.6 (#186), Trinity Large Thinking: 40.3 (#107)
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| LMArena Expert | 1258 | 1359 |
| GPQA Diamond | 60.8% | — |
| MMLU-Pro | 74% | — |
| Vectara Hallucination Rate | — | 6.9% |
| GPQA (HELM) | 59.4% | — |
Multilingual Trinity Large Thinking leads
gpt-oss-20b: 42.4 (#192), Trinity Large Thinking: 46.4 (#155)
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| LMArena Non-English | 1269 | 1325 |
| LMArena Chinese | 1311 | 1372 |
| LMArena German | 1249 | 1356 |
| LMArena Japanese | 1245 | 1312 |
| LMArena Korean | 1239 | 1304 |
| LMArena Russian | 1279 | 1338 |
| LMArena Spanish | 1267 | 1356 |
| LMArena French | — | 1374 |
Instruction Following Trinity Large Thinking leads
gpt-oss-20b: 62.0 (#235), Trinity Large Thinking: 70.7 (#157)
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| LMArena Instruction Following | 1236 | 1334 |
| IFEval | 73.2% | — |
Long Context Trinity Large Thinking leads
gpt-oss-20b: 37.9 (#203), Trinity Large Thinking: 41.2 (#139)
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| LMArena Longer Query | 1251 | 1355 |
Writing & Preference Trinity Large Thinking leads
gpt-oss-20b: 35.8 (#260), Trinity Large Thinking: 54.0 (#152)
| Benchmark | gpt-oss-20b | Trinity Large Thinking |
|---|---|---|
| LMArena Text | 1288 | 1340 |
| LMArena Creative Writing | 1201 | 1321 |
| LMArena Multi-Turn | 1268 | 1342 |
| EQ-Bench Creative Writing | 666 | — |
| WildBench | 73.7% | — |
Frequently asked questions
Is gpt-oss-20b better than Trinity Large Thinking?
Trinity Large Thinking is the stronger model overall, scoring 38.1 to 32.1 on the Noometry Index. gpt-oss-20b costs 11× less per token, which makes it the better buy when Trinity Large Thinking's lead doesn't matter for your workload.
Which is cheaper, gpt-oss-20b or Trinity Large Thinking?
gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; Trinity Large Thinking lists at $0.25 and $0.80.
Is gpt-oss-20b or Trinity Large Thinking better for coding?
gpt-oss-20b scores higher on coding benchmarks: 37.1 versus 33.8 in the Noometry coding category.
Which has the bigger context window?
Trinity Large Thinking does, with 262K tokens against 131K.
How many benchmarks do gpt-oss-20b and Trinity Large Thinking share?
18 benchmarks have published results for both models. gpt-oss-20b has 34 scored results on Noometry and Trinity Large Thinking has 24.