Model comparison
Command R vs DeepSeek V4 Pro
DeepSeek V4 Pro is the stronger model overall, scoring 54.2 to 31.0 on the Noometry Index. Command R costs 2.1× less per token, which makes it the better buy when DeepSeek V4 Pro's lead doesn't matter for your workload.
Last verified . 19 shared benchmarks.
Cohere
DeepSeekSummary
- They share 19 benchmarks with published results for both. Command R scores higher in 0 categories and DeepSeek V4 Pro in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Pro leads 56.4 to 13.3.
- The biggest single-benchmark swing is DTBench: 46.4% for Command R and 93.9% for DeepSeek V4 Pro.
- Command R is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.43 / $0.87 for DeepSeek V4 Pro.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 128K.
Side by side
| Command R | DeepSeek V4 Pro | |
|---|---|---|
| Provider | Cohere | DeepSeek |
| Noometry Index | 31.0 | 54.2 |
| Released | 2024-08-30 | 2026-04-24 |
| Weights | Open | Open |
| Context window | 128K | 1M |
| Max output | 4K | 393K |
| Input $ / M tokens | $0.15 | $0.43 |
| Output $ / M tokens | $0.60 | $0.87 |
| Results tracked | 29 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Pro leads
Command R: 28.8 (#301), DeepSeek V4 Pro: 52.2 (#31)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| LMArena Coding | 1170 | 1470 |
| SWE-bench Verified | — | 77.6% |
| FrontierCode | — | 28.6% |
| LMArena WebDev | — | 1582 |
| SciCode | — | 51% |
| WeirdML | — | 66.2% |
| BigCodeBench Instruct | 37.1% | — |
| LiveBench Coding | 17.9% | — |
| BigCodeBench Complete | 45.2% | — |
| ALE-Bench | — | 1,403 |
Agentic & Tool Use Not comparable
Command R: —, DeepSeek V4 Pro: 32.7 (#54)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| APEX-Agents | — | 47.3% |
| Vending-Bench 2 | — | 3,285 |
Reasoning DeepSeek V4 Pro leads
Command R: 13.3 (#329), DeepSeek V4 Pro: 56.4 (#22)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| LMArena Hard Prompts | 1164 | 1461 |
| DTBench | 46.4% | 93.9% |
| LMCA | 9.2% | 45.5% |
| ARC-AGI-2 | — | 61.3% |
| Kagi LLM Benchmark | — | 53.5% |
| NYT Connections (extended) | — | 91.3% |
| ARC-AGI-1 | — | 90.5% |
| CritPt | — | 18% |
| Chess Puzzles | — | 47% |
| LiveBench Reasoning | 21.9% | — |
| Mystery Game Puzzles | — | 43% |
| LiveBench Data Analysis | 33.3% | — |
| Surface Evolver Bench | — | 40% |
| Epoch Capabilities Index | — | 155.39 |
| ForecastBench | — | 56.1 |
| LiveBench | 27.5% | — |
Math DeepSeek V4 Pro leads
Command R: 25.8 (#244), DeepSeek V4 Pro: 64.6 (#24)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| LMArena Math | 1155 | 1455 |
| FrontierMath (Tiers 1-3) | — | 64.6% |
| FrontierMath Tier 4 | — | 26.8% |
| MathArena Final-Answer Competitions | — | 76.6% |
| OTIS Mock AIME 2024-2025 | — | 98.6% |
| ProofBench | — | 50% |
| LiveBench Math | 19.4% | — |
Knowledge DeepSeek V4 Pro leads
Command R: 30.5 (#219), DeepSeek V4 Pro: 59.5 (#28)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| LMArena Expert | 1138 | 1463 |
| GPQA Diamond | — | 91.7% |
| SimpleQA Verified | — | 52.9% |
| Vectara Hallucination Rate | — | 8.6% |
| MMLU | 65.2% | — |
Multilingual DeepSeek V4 Pro leads
Command R: 35.9 (#240), DeepSeek V4 Pro: 54.5 (#44)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| LMArena Non-English | 1174 | 1438 |
| LMArena Chinese | 1182 | 1486 |
| LMArena French | 1161 | 1473 |
| LMArena German | 1176 | 1458 |
| LMArena Japanese | 1143 | 1448 |
| LMArena Korean | 1162 | 1452 |
| LMArena Russian | 1174 | 1452 |
| LMArena Spanish | 1150 | 1457 |
Instruction Following DeepSeek V4 Pro leads
Command R: 58.3 (#256), DeepSeek V4 Pro: 76.3 (#44)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| LMArena Instruction Following | 1167 | 1448 |
| LiveBench Instruction Following | 55.6% | — |
Long Context DeepSeek V4 Pro leads
Command R: 36.3 (#226), DeepSeek V4 Pro: 45.0 (#49)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| LMArena Longer Query | 1198 | 1459 |
| CL-bench Life | — | 13.5% |
Writing & Preference DeepSeek V4 Pro leads
Command R: 38.4 (#248), DeepSeek V4 Pro: 65.6 (#45)
| Benchmark | Command R | DeepSeek V4 Pro |
|---|---|---|
| LMArena Text | 1187 | 1451 |
| LMArena Creative Writing | 1170 | 1446 |
| LMArena Multi-Turn | 1163 | 1467 |
| EQ-Bench Creative Writing | — | 1553 |
| EQ-Bench 4 | — | 1166 |
| LiveBench Language | 16.7% | — |
Frequently asked questions
Is Command R better than DeepSeek V4 Pro?
DeepSeek V4 Pro is the stronger model overall, scoring 54.2 to 31.0 on the Noometry Index. Command R costs 2.1× less per token, which makes it the better buy when DeepSeek V4 Pro's lead doesn't matter for your workload.
Which is cheaper, Command R or DeepSeek V4 Pro?
Command R is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; DeepSeek V4 Pro lists at $0.43 and $0.87.
Is Command R or DeepSeek V4 Pro better for coding?
DeepSeek V4 Pro scores higher on coding benchmarks: 52.2 versus 28.8 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4 Pro does, with 1M tokens against 128K.
How many benchmarks do Command R and DeepSeek V4 Pro share?
19 benchmarks have published results for both models. Command R has 29 scored results on Noometry and DeepSeek V4 Pro has 48.