Model comparison
Command R vs DeepSeek-V3
DeepSeek-V3 is the stronger model overall, scoring 39.2 to 31.0 on the Noometry Index. Command R costs 1.5× less per token, which makes it the better buy when DeepSeek-V3's lead doesn't matter for your workload.
Last verified . 29 shared benchmarks.
Cohere
DeepSeekSummary
- They share 29 benchmarks with published results for both. Command R scores higher in 1 category and DeepSeek-V3 in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3 leads 57.6 to 38.4.
- The biggest single-benchmark swing is LiveBench Math: 19.4% for Command R and 73.5% for DeepSeek-V3.
- Command R is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.24 / $0.90 for DeepSeek-V3.
- DeepSeek-V3 accepts more context: 164K tokens versus 128K.
Side by side
| Command R | DeepSeek-V3 | |
|---|---|---|
| Provider | Cohere | DeepSeek |
| Noometry Index | 31.0 | 39.2 |
| Released | 2024-08-30 | 2024-12-26 |
| Weights | Open | Open |
| Context window | 128K | 164K |
| Max output | 4K | 164K |
| Input $ / M tokens | $0.15 | $0.24 |
| Output $ / M tokens | $0.60 | $0.90 |
| Results tracked | 29 | 60 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
Command R: 28.8 (#301), DeepSeek-V3: 41.9 (#99)
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| BigCodeBench Instruct | 37.1% | 50% |
| LiveBench Coding | 17.9% | 70.9% |
| LMArena Coding | 1170 | 1368 |
| BigCodeBench Complete | 45.2% | 62.2% |
| Aider Polyglot | — | 55.1% |
| SciCode | — | 35.8% |
| WeirdML | — | 36.1% |
| HumanEval+ | — | 86.6% |
| MBPP+ | — | 73% |
Agentic & Tool Use Not comparable
Command R: —, DeepSeek-V3: —
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| METR Time Horizons | — | 49.6% |
Reasoning DeepSeek-V3 leads
Command R: 13.3 (#329), DeepSeek-V3: 20.0 (#226)
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| LiveBench Reasoning | 21.9% | 65.8% |
| LMArena Hard Prompts | 1164 | 1365 |
| DTBench | 46.4% | 64.8% |
| LiveBench Data Analysis | 33.3% | 60.9% |
| LMCA | 9.2% | 15.5% |
| LiveBench | 27.5% | 66.9% |
| SimpleBench | — | 27.2% |
| Kagi LLM Benchmark | — | 52.3% |
| CritPt | — | 0% |
| BIG-Bench Hard | — | 87.5% |
| Epoch Capabilities Index | — | 135.95 |
| ForecastBench | — | 59.1 |
| HellaSwag | — | 88.9% |
| PIQA | — | 84.7% |
| WinoGrande | — | 85.2% |
Math DeepSeek-V3 leads
Command R: 25.8 (#244), DeepSeek-V3: 30.4 (#208)
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| LiveBench Math | 19.4% | 73.5% |
| LMArena Math | 1155 | 1373 |
| OTIS Mock AIME 2024-2025 | — | 37.8% |
| Omni-MATH | — | 40.3% |
| MATH Level 5 | — | 75.5% |
| FrontierMath (Feb 2025 set) | — | 1.7% |
Knowledge DeepSeek-V3 leads
Command R: 30.5 (#219), DeepSeek-V3: 37.5 (#147)
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| LMArena Expert | 1138 | 1352 |
| MMLU | 65.2% | 87.2% |
| GPQA Diamond | — | 67.6% |
| MMLU-Pro | — | 72.3% |
| Confabulations | — | 26.1% |
| Vectara Hallucination Rate | — | 6.1% |
| GPQA (HELM) | — | 53.8% |
| ARC (AI2) Challenge | — | 95.3% |
| TriviaQA | — | 82.9% |
Multilingual DeepSeek-V3 leads
Command R: 35.9 (#240), DeepSeek-V3: 48.8 (#138)
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| LMArena Non-English | 1174 | 1358 |
| LMArena Chinese | 1182 | 1390 |
| LMArena French | 1161 | 1386 |
| LMArena German | 1176 | 1374 |
| LMArena Japanese | 1143 | 1333 |
| LMArena Korean | 1162 | 1321 |
| LMArena Russian | 1174 | 1372 |
| LMArena Spanish | 1150 | 1358 |
Instruction Following DeepSeek-V3 leads
Command R: 58.3 (#256), DeepSeek-V3: 73.0 (#125)
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| LiveBench Instruction Following | 55.6% | 81.5% |
| LMArena Instruction Following | 1167 | 1345 |
| IFEval | — | 83.2% |
Long Context Command R leads
Command R: 36.3 (#226), DeepSeek-V3: 34.0 (#248)
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| LMArena Longer Query | 1198 | 1353 |
| Fiction.LiveBench | — | 50% |
Writing & Preference DeepSeek-V3 leads
Command R: 38.4 (#248), DeepSeek-V3: 57.6 (#124)
| Benchmark | Command R | DeepSeek-V3 |
|---|---|---|
| LMArena Text | 1187 | 1375 |
| LMArena Creative Writing | 1170 | 1364 |
| LMArena Multi-Turn | 1163 | 1389 |
| LiveBench Language | 16.7% | 49.1% |
| Short-Story Creative Writing | — | 77% |
| EQ-Bench Creative Writing | — | 1472 |
| WildBench | — | 83% |
Frequently asked questions
Is Command R better than DeepSeek-V3?
DeepSeek-V3 is the stronger model overall, scoring 39.2 to 31.0 on the Noometry Index. Command R costs 1.5× less per token, which makes it the better buy when DeepSeek-V3's lead doesn't matter for your workload.
Which is cheaper, Command R or DeepSeek-V3?
Command R is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; DeepSeek-V3 lists at $0.24 and $0.90.
Is Command R or DeepSeek-V3 better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 41.9 versus 28.8 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3 does, with 164K tokens against 128K.
How many benchmarks do Command R and DeepSeek-V3 share?
29 benchmarks have published results for both models. Command R has 29 scored results on Noometry and DeepSeek-V3 has 60.