# Command A vs Step 3

> Step 3 is the stronger model overall, scoring 39.9 to 35.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/command-a-vs-step-3
- Last updated: 2026-10-10
- Shared benchmarks: 16

## Summary

- They share 16 benchmarks with published results for both. Command A scores higher in 2 categories and Step 3 in 6 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in coding, where Step 3 leads 39.5 to 26.8.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 28.8% for Command A and 62.3% for Step 3.

## Snapshot

| | Command A | Step 3 |
|---|---|---|
| Provider | Cohere | StepFun |
| Noometry Index | 35.9 | 39.9 |
| Rank | 212 | 145 |
| Context | 256K | — |
| Input $/M | $2.50 | — |
| Output $/M | $10 | — |
| Weights | Open | Open |

## Coding

- Command A: 26.8 (#317)
- Step 3: 39.5 (#144)

| Benchmark | Command A | Step 3 |
|---|---|---|
| LMArena Coding | 1331 | 1367 |
| Aider Polyglot | 12% | — |

## Agentic & Tool Use

- Command A: 35.7 (#36)
- Step 3: —

| Benchmark | Command A | Step 3 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 57.1% | — |

## Reasoning

- Command A: 17.7 (#276)
- Step 3: 27.6 (#102)

| Benchmark | Command A | Step 3 |
|---|---|---|
| Kagi LLM Benchmark | 28.8% | 62.3% |
| LMArena Hard Prompts | 1326 | 1355 |
| DTBench | 61.3% | — |
| LMCA | 10.3% | — |

## Math

- Command A: 33.5 (#172)
- Step 3: 34.9 (#148)

| Benchmark | Command A | Step 3 |
|---|---|---|
| LMArena Math | 1300 | 1366 |

## Knowledge

- Command A: 36.5 (#157)
- Step 3: 36.2 (#161)

| Benchmark | Command A | Step 3 |
|---|---|---|
| LMArena Expert | 1295 | 1333 |
| Vectara Hallucination Rate | 9.3% | — |

## Multimodal

- Command A: —
- Step 3: 34.9 (#81)

| Benchmark | Command A | Step 3 |
|---|---|---|
| LMArena Vision | — | 1176 |

## Multilingual

- Command A: 45.5 (#166)
- Step 3: 46.5 (#154)

| Benchmark | Command A | Step 3 |
|---|---|---|
| LMArena Non-English | 1313 | 1327 |
| LMArena Chinese | 1327 | 1396 |
| LMArena German | 1341 | 1371 |
| LMArena Korean | 1285 | 1271 |
| LMArena Russian | 1314 | 1331 |
| LMArena Spanish | 1346 | 1372 |
| LMArena French | 1352 | — |
| LMArena Japanese | 1283 | — |

## Instruction Following

- Command A: 69.3 (#172)
- Step 3: 70.6 (#159)

| Benchmark | Command A | Step 3 |
|---|---|---|
| LMArena Instruction Following | 1309 | 1332 |

## Long Context

- Command A: 40.6 (#147)
- Step 3: 40.3 (#152)

| Benchmark | Command A | Step 3 |
|---|---|---|
| LMArena Longer Query | 1334 | 1326 |

## Writing & Preference

- Command A: 47.8 (#204)
- Step 3: 54.5 (#146)

| Benchmark | Command A | Step 3 |
|---|---|---|
| LMArena Text | 1331 | 1351 |
| LMArena Creative Writing | 1319 | 1322 |
| LMArena Multi-Turn | 1338 | 1340 |
| EQ-Bench Creative Writing | 1145 | — |

## FAQ

### Is Command A better than Step 3?

Step 3 is the stronger model overall, scoring 39.9 to 35.9 on the Noometry Index.

### Is Command A or Step 3 better for coding?

Step 3 scores higher on coding benchmarks: 39.5 versus 26.8 in the Noometry coding category.

### How many benchmarks do Command A and Step 3 share?

16 benchmarks have published results for both models. Command A has 24 scored results on Noometry and Step 3 has 17.