Model comparison
Command R vs GPT-5.4 nano
GPT-5.4 nano is the stronger model overall, scoring 41.9 to 31.4 on the Noometry Index. Command R costs 1.8× less per token, which makes it the better buy when GPT-5.4 nano's lead doesn't matter for your workload.
Last verified . 19 shared benchmarks.
CohereSummary
- They share 19 benchmarks with published results for both. Command R scores higher in 0 categories and GPT-5.4 nano in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where GPT-5.4 nano leads 55.7 to 38.2.
- The biggest single-benchmark swing is DTBench: 46.4% for Command R and 80.3% for GPT-5.4 nano.
- Command R is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.20 / $1.25 for GPT-5.4 nano.
- GPT-5.4 nano accepts more context: 400K tokens versus 128K.
- Command R has downloadable open weights; the other is API-only.
Side by side
| Command R | GPT-5.4 nano | |
|---|---|---|
| Provider | Cohere | OpenAI |
| Noometry Index | 31.4 | 41.9 |
| Released | 2024-08-30 | 2026-03-17 |
| Weights | Open | Proprietary |
| Context window | 128K | 400K |
| Max output | 4K | 128K |
| Input $ / M tokens | $0.15 | $0.20 |
| Output $ / M tokens | $0.60 | $1.25 |
| Results tracked | 29 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 nano leads
Command R: 29.3 (#306), GPT-5.4 nano: 43.6 (#84)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Coding | 1169 | 1405 |
| SciCode | — | 46.9% |
| WeirdML | — | 49.2% |
| BigCodeBench Instruct | 37.1% | — |
| LiveBench Coding | 17.9% | — |
| BigCodeBench Complete | 45.2% | — |
| ALE-Bench | — | 1,005 |
Reasoning GPT-5.4 nano leads
Command R: 13.8 (#331), GPT-5.4 nano: 23.7 (#173)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Hard Prompts | 1164 | 1381 |
| DTBench | 46.4% | 80.3% |
| LMCA | 9.2% | 36.9% |
| ARC-AGI-2 | — | 5.7% |
| Kagi LLM Benchmark | — | 39.7% |
| ARC-AGI-1 | — | 51.5% |
| CritPt | — | 9.3% |
| Chess Puzzles | — | 30% |
| LiveBench Reasoning | 21.9% | — |
| Mystery Game Puzzles | — | 9% |
| LiveBench Data Analysis | 33.3% | — |
| Epoch Capabilities Index | — | 145.81 |
| ForecastBench | — | 57.3 |
| LiveBench | 27.5% | — |
Math GPT-5.4 nano leads
Command R: 28.0 (#246), GPT-5.4 nano: 40.9 (#88)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Math | 1155 | 1406 |
| FrontierMath (Tiers 1-3) | — | 44.9% |
| FrontierMath Tier 4 | — | 12.2% |
| OTIS Mock AIME 2024-2025 | — | 87.8% |
| ProofBench | — | 5% |
| LiveBench Math | 19.4% | — |
| FrontierMath (Feb 2025 set) | — | 25.9% |
| FrontierMath Tier 4 (v1) | — | 6.3% |
Knowledge GPT-5.4 nano leads
Command R: 31.0 (#221), GPT-5.4 nano: 41.9 (#103)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Expert | 1138 | 1396 |
| GPQA Diamond | — | 78.5% |
| SimpleQA Verified | — | 11.7% |
| Vectara Hallucination Rate | — | 3.1% |
| MMLU | 65.2% | — |
Multimodal Not comparable
Command R: —, GPT-5.4 nano: 36.7 (#78)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Vision | — | 1196 |
Multilingual GPT-5.4 nano leads
Command R: 35.7 (#245), GPT-5.4 nano: 48.6 (#140)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Non-English | 1174 | 1359 |
| LMArena Chinese | 1182 | 1392 |
| LMArena French | 1162 | 1396 |
| LMArena German | 1176 | 1367 |
| LMArena Japanese | 1143 | 1343 |
| LMArena Korean | 1163 | 1320 |
| LMArena Russian | 1174 | 1363 |
| LMArena Spanish | 1151 | 1371 |
Instruction Following GPT-5.4 nano leads
Command R: 58.1 (#261), GPT-5.4 nano: 71.9 (#144)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Instruction Following | 1167 | 1362 |
| LiveBench Instruction Following | 55.6% | — |
Long Context GPT-5.4 nano leads
Command R: 36.3 (#231), GPT-5.4 nano: 41.6 (#137)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Longer Query | 1198 | 1366 |
Writing & Preference GPT-5.4 nano leads
Command R: 38.2 (#254), GPT-5.4 nano: 55.7 (#142)
| Benchmark | Command R | GPT-5.4 nano |
|---|---|---|
| LMArena Text | 1187 | 1372 |
| LMArena Creative Writing | 1170 | 1314 |
| LMArena Multi-Turn | 1163 | 1382 |
| LiveBench Language | 16.7% | — |
Frequently asked questions
Is Command R better than GPT-5.4 nano?
GPT-5.4 nano is the stronger model overall, scoring 41.9 to 31.4 on the Noometry Index. Command R costs 1.8× less per token, which makes it the better buy when GPT-5.4 nano's lead doesn't matter for your workload.
Which is cheaper, Command R or GPT-5.4 nano?
Command R is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; GPT-5.4 nano lists at $0.20 and $1.25.
Is Command R or GPT-5.4 nano better for coding?
GPT-5.4 nano scores higher on coding benchmarks: 43.6 versus 29.3 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 nano does, with 400K tokens against 128K.
How many benchmarks do Command R and GPT-5.4 nano share?
19 benchmarks have published results for both models. Command R has 29 scored results on Noometry and GPT-5.4 nano has 40.