Cohereproprietary
Released: 2024-04-04Command R+
Cohere 104B enterprise-grade model optimized for multi-step retrieval-augmented generation (RAG) and tool use.
Intelligence Index
74.0/ 100
Calibrated multi-domain compositeThroughput Speed
50tok/sec
Output streaming throughputTime To First Token
480ms
Initial chunk server latencyLiveBench Contam-Free
—%
September 2026 suite scoreSourced Benchmark Evaluations (22)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| StructFlowBench | Nearing Saturation | 57.9% | 75 | 2024-04-22 | independent | Source ↗ |
| FireBench | Active | 57.2% | 95 | 2024-04-20 | independent | Source ↗ |
| SimpleQA | Active | 38.5% | 90 | 2024-04-19 | independent | Source ↗ |
| OpenAI-MRCR | Active | 62.2% | 80 | 2024-04-19 | independent | Source ↗ |
| ComplexBench | Nearing Saturation | 55.8% | 80 | 2024-04-18 | independent | Source ↗ |
| LongBench-Pro | Active | 60.7% | 80 | 2024-04-18 | independent | Source ↗ |
| ZeroSCROLLS | Nearing Saturation | 59.2% | 78 | 2024-04-17 | independent | Source ↗ |
| InfiniteBench | Nearing Saturation | 62.9% | 65 | 2024-04-16 | independent | Source ↗ |
| AgentIF | Nearing Saturation | 46.3% | 80 | 2024-04-16 | independent | Source ↗ |
| MMLU-Pro | Nearing Saturation | 68.1% | 75 | 2024-04-15 | independent | Source ↗ |
| SCROLLS | Nearing Saturation | 59.2% | 65 | 2024-04-15 | independent | Source ↗ |
| RULER | Nearing Saturation | 68.1% | 84 | 2024-04-14 | independent | Source ↗ |
| MMLU-Redux | Saturated | 72.5% | 25 | 2024-04-13 | independent | Source ↗ |
| QuALITY | Nearing Saturation | 65.1% | 78 | 2024-04-13 | independent | Source ↗ |
| LongBench | Nearing Saturation | 63.6% | 65 | 2024-04-12 | independent | Source ↗ |
| QASPER | Nearing Saturation | 61.4% | 65 | 2024-04-11 | independent | Source ↗ |
| TruthfulQA | Saturated | 65.1% | 25 | 2024-04-11 | independent | Source ↗ |
| MT-Bench | Nearing Saturation | 7% | 75 | 2024-04-09 | independent | Source ↗ |
| IFEval | Deprecated | 66.7% | 25 | 2024-04-08 | independent | Source ↗ |
| GPQA | Nearing Saturation | 65.1% | 78 | 2024-04-08 | independent | Source ↗ |
| Needle-in-a-Haystack | Saturated | 94.8% | 44 | 2024-04-07 | independent | Source ↗ |
| MMLU | Saturated | 75.5% | 25 | 2024-04-05 | independent | Source ↗ |
OpenAI Compatible API ExecutionAPI Ready
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="command-r-plus",
messages=[{"role": "user", "content": "Evaluate multi-step logic problem"}],
temperature=0.2
)
print(response.choices[0].message.content)Evaluation & Sourcing Notes
Scores listed for Command R+ represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.