vs
2:0
overall capability wins (0 ties across 2 scored). Command A Reasoning leads overall.
- Pricing0:3
- Speed0:2
- Specs2:0
Verdict
Command A Reasoning offers ~62.5x larger context (256K vs 4K); Command A Reasoning allows ~8.0x more max output (33K vs 4K); Command Light is ~11.7x cheaper on a blended token basis than Command A Reasoning.
- Command A Reasoning offers ~62.5x larger context (256K vs 4K)
- Command A Reasoning allows ~8.0x more max output (33K vs 4K)
- Command Light is ~11.7x cheaper on a blended token basis than Command A Reasoning
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Command A Reasoning
- $5
- Command Light
- $0.45
Command Light is about 11.1x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Cohere
Cohere
Family
Command
Command
License
Proprietary
Proprietary
Open weights
No
No
Release
Aug 1, 2025
Nov 2, 2023
Knowledge cutoff
-
-
API / provider
Cohere
Cohere
Modalities
text → text
text → text
Specs
Context window
256K
4K
Max output
33K
4K
Parameters
—
—
Pricing
Input $/1M
$2.50
$0.30
Output $/1M
$10
$0.60
Blended $/1M (3∶1)
$4.38
$0.38
Speed
tok/s
45
120
TTFT (s)
0.8
0.2
Reasoning
MMLU-Pro
—
—
GPQA Diamond
—
—
Humanity's Last Exam
—
—
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
—
—
AA-LCR v1.1
—
—
CritPt
—
—
MMMU-Pro
—
—
IFBench
—
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
—
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
—
—
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
—
| Metric | Command A Reasoning | Command Light | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Cohere | Cohere | — |
| Family | Command | Command | — |
| License | Proprietary | Proprietary | — |
| Open weights | No | No | — |
| Release | Aug 1, 2025 | Nov 2, 2023 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Cohere | Cohere | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 256K | 4K | A +252K |
| Max output | 33K | 4K | A +29K |
| Parameters | — | — | — |
| Pricing | |||
| Input $/1M | $2.50 | $0.30 | B +$2.20 |
| Output $/1M | $10 | $0.60 | B +$9.40 |
| Blended $/1M (3∶1) | $4.38 | $0.38 | B +$4 |
| Speed | |||
| tok/s | 45 | 120 | B +75 |
| TTFT (s) | 0.8 | 0.2 | B +0.60 |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | — | — | — |
| Humanity's Last Exam | — | — | — |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | — | — | — |
| AA-LCR v1.1 | — | — | — |
| CritPt | — | — | — |
| MMMU-Pro | — | — | — |
| IFBench | — | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | — | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | — | — | — |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
Command A Reasoning
- Context
- 256K / 33K out
- Parameters
- —
- Price
- $2.50 / $10
- Speed
- 45 tok/s · 0.8s TTFT
- Modalities
- text → text
- License
- Proprietary
Cohere's first explicit reasoning model — thinks before answering for nuanced agent and STEM workflows at 256K context.
Command Light
- Context
- 4K / 4K out
- Parameters
- —
- Price
- $0.30 / $0.60
- Speed
- 120 tok/s · 0.2s TTFT
- Modalities
- text → text
- License
- Proprietary
Cohere's budget Command SKU for classification and short RAG answers.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.
These two models have no overlapping published benchmarks in our dataset. Compare specs and pricing instead.