# Step 3.7 Flash vs Trinity Large Thinking

> Trinity Large Thinking is the stronger model overall, scoring 38.1 to 36.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/step-3-7-flash-vs-trinity-large-thinking
- Last updated: 2026-10-10
- Shared benchmarks: 3

## Summary

- They share 3 benchmarks with published results for both. Step 3.7 Flash scores higher in 3 categories and Trinity Large Thinking in 0 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in math, where Step 3.7 Flash leads 40.9 to 34.9.
- The biggest single-benchmark swing is NYT Connections (extended): 39.7% for Step 3.7 Flash and 16.5% for Trinity Large Thinking.
- Trinity Large Thinking is cheaper at $0.25 / $0.80 per million input/output tokens, against $0.18 / $1.11 for Step 3.7 Flash.
- Trinity Large Thinking accepts more context: 262K tokens versus 256K.

## Snapshot

| | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| Provider | StepFun | Arcee AI |
| Noometry Index | 36.5 | 38.1 |
| Rank | 207 | 180 |
| Context | 256K | 262K |
| Input $/M | $0.18 | $0.25 |
| Output $/M | $1.11 | $0.80 |
| Weights | Open | Open |

## Coding

- Step 3.7 Flash: 39.4 (#147)
- Trinity Large Thinking: 33.8 (#238)

| Benchmark | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| SciCode | 40% | 36.1% |
| LMArena WebDev | — | 1238 |
| LMArena Coding | — | 1381 |
| ALE-Bench | 694.12 | — |

## Reasoning

- Step 3.7 Flash: 20.9 (#211)
- Trinity Large Thinking: 16.3 (#298)

| Benchmark | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| NYT Connections (extended) | 39.7% | 16.5% |
| CritPt | 2.3% | 0.9% |
| Thematic Generalization | — | 41.6% |
| LMArena Hard Prompts | — | 1350 |
| Surface Evolver Bench | — | 15.6% |

## Math

- Step 3.7 Flash: 40.9 (#74)
- Trinity Large Thinking: 34.9 (#149)

| Benchmark | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| MathArena Final-Answer Competitions | 68.5% | — |
| LMArena Math | — | 1366 |

## Knowledge

- Step 3.7 Flash: —
- Trinity Large Thinking: 40.3 (#107)

| Benchmark | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| Vectara Hallucination Rate | — | 6.9% |
| LMArena Expert | — | 1359 |

## Multilingual

- Step 3.7 Flash: —
- Trinity Large Thinking: 46.4 (#155)

| Benchmark | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Non-English | — | 1325 |
| LMArena Chinese | — | 1372 |
| LMArena French | — | 1374 |
| LMArena German | — | 1356 |
| LMArena Japanese | — | 1312 |
| LMArena Korean | — | 1304 |
| LMArena Russian | — | 1338 |
| LMArena Spanish | — | 1356 |

## Instruction Following

- Step 3.7 Flash: —
- Trinity Large Thinking: 70.7 (#157)

| Benchmark | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Instruction Following | — | 1334 |

## Long Context

- Step 3.7 Flash: —
- Trinity Large Thinking: 41.2 (#139)

| Benchmark | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Longer Query | — | 1355 |

## Writing & Preference

- Step 3.7 Flash: —
- Trinity Large Thinking: 54.0 (#152)

| Benchmark | Step 3.7 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Text | — | 1340 |
| LMArena Creative Writing | — | 1321 |
| LMArena Multi-Turn | — | 1342 |

## FAQ

### Is Step 3.7 Flash better than Trinity Large Thinking?

Trinity Large Thinking is the stronger model overall, scoring 38.1 to 36.5 on the Noometry Index.

### Which is cheaper, Step 3.7 Flash or Trinity Large Thinking?

Trinity Large Thinking is cheaper. It lists at $0.25 per million input tokens and $0.80 per million output tokens; Step 3.7 Flash lists at $0.18 and $1.11.

### Is Step 3.7 Flash or Trinity Large Thinking better for coding?

Step 3.7 Flash scores higher on coding benchmarks: 39.4 versus 33.8 in the Noometry coding category.

### Which has the bigger context window?

Trinity Large Thinking does, with 262K tokens against 256K.

### How many benchmarks do Step 3.7 Flash and Trinity Large Thinking share?

3 benchmarks have published results for both models. Step 3.7 Flash has 5 scored results on Noometry and Trinity Large Thinking has 24.