Model comparison
C4ai Aya Expanse 32b vs DeepSeek-V3
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 35.9 on the Noometry Index.
Last verified . 18 shared benchmarks.
Cohere
DeepSeekSummary
- They share 18 benchmarks with published results for both. C4ai Aya Expanse 32b scores higher in 3 categories and DeepSeek-V3 in 5 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3 leads 57.4 to 42.2.
- DeepSeek-V3 accepts more context: 164K tokens versus 128K.
Side by side
| C4ai Aya Expanse 32b | DeepSeek-V3 | |
|---|---|---|
| Provider | Cohere | DeepSeek |
| Noometry Index | 35.9 | 39.5 |
| Released | 2024-10-24 | 2024-12-26 |
| Weights | Open | Open |
| Context window | 128K | 164K |
| Max output | 4K | 164K |
| Input $ / M tokens | — | $0.24 |
| Output $ / M tokens | — | $0.90 |
| Results tracked | 18 | 60 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
C4ai Aya Expanse 32b: 34.8 (#231), DeepSeek-V3: 42.3 (#106)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| LMArena Coding | 1197 | 1368 |
| Aider Polyglot | — | 55.1% |
| SciCode | — | 35.8% |
| WeirdML | — | 36.1% |
| BigCodeBench Instruct | — | 50% |
| LiveBench Coding | — | 70.9% |
| BigCodeBench Complete | — | 62.2% |
| HumanEval+ | — | 86.6% |
| MBPP+ | — | 73% |
Agentic & Tool Use Not comparable
C4ai Aya Expanse 32b: —, DeepSeek-V3: —
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| METR Time Horizons | — | 49.6% |
Reasoning C4ai Aya Expanse 32b leads
C4ai Aya Expanse 32b: 23.3 (#180), DeepSeek-V3: 20.5 (#236)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| LMArena Hard Prompts | 1193 | 1365 |
| SimpleBench | — | 27.2% |
| Kagi LLM Benchmark | — | 52.3% |
| CritPt | — | 0% |
| LiveBench Reasoning | — | 65.8% |
| DTBench | — | 64.8% |
| LiveBench Data Analysis | — | 60.9% |
| LMCA | — | 15.5% |
| BIG-Bench Hard | — | 87.5% |
| Epoch Capabilities Index | — | 135.94 |
| ForecastBench | — | 59.1 |
| HellaSwag | — | 88.9% |
| LiveBench | — | 66.9% |
| PIQA | — | 84.7% |
| WinoGrande | — | 85.2% |
Math C4ai Aya Expanse 32b leads
C4ai Aya Expanse 32b: 34.0 (#197), DeepSeek-V3: 32.1 (#219)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| LMArena Math | 1200 | 1373 |
| OTIS Mock AIME 2024-2025 | — | 37.8% |
| Omni-MATH | — | 40.3% |
| LiveBench Math | — | 73.5% |
| MATH Level 5 | — | 75.5% |
| FrontierMath (Feb 2025 set) | — | 1.7% |
Knowledge DeepSeek-V3 leads
C4ai Aya Expanse 32b: 33.2 (#206), DeepSeek-V3: 37.5 (#155)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| Vectara Hallucination Rate | 10.9% | 6.1% |
| LMArena Expert | 1182 | 1351 |
| GPQA Diamond | — | 67.6% |
| MMLU-Pro | — | 72.3% |
| Confabulations | — | 26.1% |
| GPQA (HELM) | — | 53.8% |
| ARC (AI2) Challenge | — | 95.3% |
| MMLU | — | 87.2% |
| TriviaQA | — | 82.9% |
Multilingual DeepSeek-V3 leads
C4ai Aya Expanse 32b: 38.4 (#230), DeepSeek-V3: 48.5 (#143)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| LMArena Non-English | 1213 | 1358 |
| LMArena Chinese | 1211 | 1391 |
| LMArena French | 1248 | 1385 |
| LMArena German | 1199 | 1374 |
| LMArena Japanese | 1163 | 1333 |
| LMArena Korean | 1158 | 1319 |
| LMArena Russian | 1227 | 1373 |
| LMArena Spanish | 1193 | 1358 |
Instruction Following DeepSeek-V3 leads
C4ai Aya Expanse 32b: 62.6 (#237), DeepSeek-V3: 72.8 (#130)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| LMArena Instruction Following | 1196 | 1345 |
| LiveBench Instruction Following | — | 81.5% |
| IFEval | — | 83.2% |
Long Context C4ai Aya Expanse 32b leads
C4ai Aya Expanse 32b: 37.2 (#220), DeepSeek-V3: 34.0 (#253)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| LMArena Longer Query | 1228 | 1352 |
| Fiction.LiveBench | — | 50% |
Writing & Preference DeepSeek-V3 leads
C4ai Aya Expanse 32b: 42.2 (#235), DeepSeek-V3: 57.4 (#130)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V3 |
|---|---|---|
| LMArena Text | 1224 | 1375 |
| LMArena Creative Writing | 1200 | 1364 |
| LMArena Multi-Turn | 1190 | 1389 |
| Short-Story Creative Writing | — | 77% |
| EQ-Bench Creative Writing | — | 1472 |
| WildBench | — | 83% |
| LiveBench Language | — | 49.1% |
Frequently asked questions
Is C4ai Aya Expanse 32b better than DeepSeek-V3?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 35.9 on the Noometry Index.
Is C4ai Aya Expanse 32b or DeepSeek-V3 better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 34.8 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3 does, with 164K tokens against 128K.
How many benchmarks do C4ai Aya Expanse 32b and DeepSeek-V3 share?
18 benchmarks have published results for both models. C4ai Aya Expanse 32b has 18 scored results on Noometry and DeepSeek-V3 has 60.