71.2%
MMLU-Pro (AA run)
Modelle / Cohere
Command · Release 2026-03-15
Cohere's Enterprise-fokussiertes Command-Tier. Stark bei RAG, Tool-Use und mehrsprachigen Retrieval-Workflows. 256K-Kontext.
Wahl, wenn der Stack bereits Cohere Embed + Rerank nutzt und ein Anbieter für Retrieval-Pipelines reicht.
48average across 7 tested families
Wo Command A auf jeder öffentlichen Benchmark-Quelle mit Zeile steht. Rang zählt jedes Modell mit neuester Zeile im selben Test — kein universeller Qualitätsscore.
| Benchmark | Family | Rang | Score | Quelle | Datum |
|---|---|---|---|---|---|
| LiveCodeBench | Coding | #12/ 13 | 29 | Artificial Analysis | 2026-07-21 |
| MATH | Math | #14/ 20 | 82 | Artificial Analysis | 2026-07-21 |
| AIME | Math | #18/ 19 | 13 | Artificial Analysis | 2026-07-21 |
| MMLU-Pro | Knowledge | #20/ 21 | 71 | Artificial Analysis | 2026-07-21 |
| τ³-Bench Banking | Agentic | #46/ 64 | 15 | Artificial Analysis | 2026-07-21 |
| AA output speed | Performance | #47/ 69 | Artificial Analysis | 2026-07-21 | |
| AA time to first token | Performance | #57/ 69 | Artificial Analysis | 2026-07-21 | |
| SciCode | Coding | #61/ 65 | 28 | Artificial Analysis | 2026-07-21 |
| AA-LCR | Long context | #62/ 64 | 18 | Artificial Analysis | 2026-07-21 |
| Humanity's Last Exam | Reasoning | #62/ 65 | 5 | Artificial Analysis | 2026-07-21 |
| Artificial Analysis Intelligence Index | Reasoning | #65/ 67 | 8 | Artificial Analysis | 2026-07-21 |
| GPQA Diamond | Reasoning | #66/ 69 | 53 | Artificial Analysis | 2026-07-21 |
| Terminal-Bench | Agentic | #68/ 69 | 1 | Artificial Analysis | 2026-07-21 |
Die drei höchstscorierenden kartierten Modelle pro Capability-Family. Wo Command A auftaucht, ist es hervorgehoben.
Benchmark-Zeilen im öffentlichen Ledger für Command A in den letzten 120 Tagen. Ältere Zeilen stehen in der vollen Tabelle unten.
| Benchmark | Family | Score | Quelle | Tage her |
|---|---|---|---|---|
| AA-LCR | Long context | 18% | Artificial Analysis | 2 |
| AA output speed | Performance | 61 t/s | Artificial Analysis | 2 |
| AA time to first token | Performance | 0.36s | Artificial Analysis | 2 |
| AIME | Math | 13% | Artificial Analysis | 2 |
| Artificial Analysis Intelligence Index | Reasoning | 7.7% | Artificial Analysis | 2 |
| GPQA Diamond | Reasoning | 52.7% | Artificial Analysis | 2 |
| Humanity's Last Exam | Reasoning | 4.6% | Artificial Analysis | 2 |
| LiveCodeBench | Coding | 28.7% | Artificial Analysis | 2 |
| MATH | Math | 81.87% | Artificial Analysis | 2 |
| MMLU-Pro | Knowledge | 71.2% | Artificial Analysis | 2 |
| SciCode | Coding | 28.1% | Artificial Analysis | 2 |
| τ³-Bench Banking | Agentic | 15.2% | Artificial Analysis | 2 |
| Terminal-Bench | Agentic | 0.76% | Artificial Analysis | 2 |
| BFCL | Tool use | 70.6% | Berkeley Function Calling Leaderboard V4 | 99 |
Ein hoher Composite, der eine schwache Family versteckt, ist eine Falle. Diese Balken zeigen Families ohne öffentlichen Test — und wo das Modell führt.
Hover or tab any dot for name, score, and input price.
Jeder Katalog-Benchmark für Command A. Scores verlinken zur Originalquelle; Lücken heißen: noch keine öffentliche Zeile.
14 of 32 catalog benchmarks have a sourced row for Command A.
71.2%
MMLU-Pro (AA run)
7.7%
AA Intelligence Index v4.1
52.7%
GPQA Diamond (AA run)
4.6%
HLE (AA run)
28.7%
LiveCodeBench (AA run)
28.1%
SciCode (AA run)
0.76%
Terminal-Bench Hard (AA run)
15.2%
τ²-Bench Banking (AA run; maps to τ³ explainer)
18%
AA-LCR (AA run)
70.6%
BFCL
61 t/s
Median output tokens/s (1k prompt, default provider)
0.36s
Median time to first token (1k prompt, default provider)