Model comparison
C4ai Aya Expanse 32b vs DeepSeek-V2.5 (Sep 2024)
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 35.9 on the Noometry Index.
Last verified . 17 shared benchmarks.
Cohere
DeepSeekSummary
- They share 17 benchmarks with published results for both. C4ai Aya Expanse 32b scores higher in 1 category and DeepSeek-V2.5 (Sep 2024) in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V2.5 (Sep 2024) leads 49.8 to 42.2.
Side by side
| C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) | |
|---|---|---|
| Provider | Cohere | DeepSeek |
| Noometry Index | 35.9 | 37.6 |
| Released | 2024-10-24 | 2024-09-06 |
| Weights | Open | Open |
| Context window | 128K | — |
| Max output | 4K | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 18 | 22 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding C4ai Aya Expanse 32b leads
C4ai Aya Expanse 32b: 34.8 (#231), DeepSeek-V2.5 (Sep 2024): 31.7 (#281)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Coding | 1197 | 1309 |
| Aider Polyglot | — | 17.8% |
| BigCodeBench Instruct | — | 48.6% |
| BigCodeBench Complete | — | 53.2% |
| HumanEval+ | — | 83.5% |
| MBPP+ | — | 74.1% |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
C4ai Aya Expanse 32b: 23.3 (#180), DeepSeek-V2.5 (Sep 2024): 25.6 (#145)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Hard Prompts | 1193 | 1289 |
Math DeepSeek-V2.5 (Sep 2024) leads
C4ai Aya Expanse 32b: 34.0 (#197), DeepSeek-V2.5 (Sep 2024): 35.9 (#177)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Math | 1200 | 1288 |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
C4ai Aya Expanse 32b: 33.2 (#206), DeepSeek-V2.5 (Sep 2024): 34.8 (#193)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Expert | 1182 | 1266 |
| Vectara Hallucination Rate | 10.9% | — |
Multilingual DeepSeek-V2.5 (Sep 2024) leads
C4ai Aya Expanse 32b: 38.4 (#230), DeepSeek-V2.5 (Sep 2024): 42.5 (#193)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Non-English | 1213 | 1273 |
| LMArena Chinese | 1211 | 1318 |
| LMArena French | 1248 | 1289 |
| LMArena German | 1199 | 1258 |
| LMArena Japanese | 1163 | 1228 |
| LMArena Korean | 1158 | 1209 |
| LMArena Russian | 1227 | 1289 |
| LMArena Spanish | 1193 | 1248 |
Instruction Following DeepSeek-V2.5 (Sep 2024) leads
C4ai Aya Expanse 32b: 62.6 (#237), DeepSeek-V2.5 (Sep 2024): 67.5 (#194)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Instruction Following | 1196 | 1280 |
Long Context DeepSeek-V2.5 (Sep 2024) leads
C4ai Aya Expanse 32b: 37.2 (#220), DeepSeek-V2.5 (Sep 2024): 39.5 (#174)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Longer Query | 1228 | 1301 |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
C4ai Aya Expanse 32b: 42.2 (#235), DeepSeek-V2.5 (Sep 2024): 49.8 (#187)
| Benchmark | C4ai Aya Expanse 32b | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Text | 1224 | 1294 |
| LMArena Creative Writing | 1200 | 1285 |
| LMArena Multi-Turn | 1190 | 1297 |
Frequently asked questions
Is C4ai Aya Expanse 32b better than DeepSeek-V2.5 (Sep 2024)?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 35.9 on the Noometry Index.
Is C4ai Aya Expanse 32b or DeepSeek-V2.5 (Sep 2024) better for coding?
C4ai Aya Expanse 32b scores higher on coding benchmarks: 34.8 versus 31.7 in the Noometry coding category.
How many benchmarks do C4ai Aya Expanse 32b and DeepSeek-V2.5 (Sep 2024) share?
17 benchmarks have published results for both models. C4ai Aya Expanse 32b has 18 scored results on Noometry and DeepSeek-V2.5 (Sep 2024) has 22.