Model comparison
DeepSeek-V2.5 (Sep 2024) vs Mistral Small
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 33.4 on the Noometry Index.
Last verified . 19 shared benchmarks.
DeepSeek
Mistral AISummary
- They share 19 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 4 categories and Mistral Small in 4 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V2.5 (Sep 2024) leads 35.9 to 16.4.
- The biggest single-benchmark swing is BigCodeBench Instruct: 48.6% for DeepSeek-V2.5 (Sep 2024) and 36.1% for Mistral Small.
Side by side
| DeepSeek-V2.5 (Sep 2024) | Mistral Small | |
|---|---|---|
| Provider | DeepSeek | Mistral AI |
| Noometry Index | 37.6 | 33.4 |
| Released | 2024-09-06 | 2024-02-26 |
| Weights | Open | Open |
| Context window | — | 262K |
| Max output | — | 256K |
| Input $ / M tokens | — | $0.15 |
| Output $ / M tokens | — | $0.60 |
| Results tracked | 22 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mistral Small leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), Mistral Small: 34.0 (#247)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| BigCodeBench Instruct | 48.6% | 36.1% |
| LMArena Coding | 1309 | 1362 |
| BigCodeBench Complete | 53.2% | 46.6% |
| Aider Polyglot | 17.8% | — |
| SciCode | — | 26.5% |
| LiveBench Coding | — | 36.2% |
| ALE-Bench | — | 497.62 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Agentic & Tool Use Not comparable
DeepSeek-V2.5 (Sep 2024): —, Mistral Small: 28.1 (#93)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 37.1% |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), Mistral Small: 19.8 (#250)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1335 |
| Kagi LLM Benchmark | — | 37.8% |
| CritPt | — | 0% |
| LiveBench Reasoning | — | 44.8% |
| DTBench | — | 70.9% |
| LiveBench Data Analysis | — | 53.7% |
| LMCA | — | 20.6% |
| LiveBench | — | 44% |
Math DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), Mistral Small: 16.4 (#293)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| LMArena Math | 1288 | 1341 |
| OTIS Mock AIME 2024-2025 | — | 5.8% |
| LiveBench Math | — | 39.9% |
| MATH Level 5 | — | 46.8% |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), Mistral Small: 31.0 (#222)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| LMArena Expert | 1266 | 1291 |
| GPQA Diamond | — | 47.5% |
| Vectara Hallucination Rate | — | 5.1% |
| MMLU | — | 68.7% |
Multimodal Not comparable
DeepSeek-V2.5 (Sep 2024): —, Mistral Small: 33.5 (#96)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| LMArena Vision | — | 1142 |
Multilingual Mistral Small leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), Mistral Small: 45.5 (#169)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| LMArena Non-English | 1273 | 1315 |
| LMArena Chinese | 1318 | 1340 |
| LMArena French | 1289 | 1337 |
| LMArena German | 1258 | 1340 |
| LMArena Japanese | 1228 | 1275 |
| LMArena Korean | 1209 | 1259 |
| LMArena Russian | 1289 | 1324 |
| LMArena Spanish | 1248 | 1346 |
Instruction Following DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), Mistral Small: 66.4 (#209)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| LMArena Instruction Following | 1280 | 1310 |
| LiveBench Instruction Following | — | 63.7% |
Long Context Too close to call
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), Mistral Small: 40.4 (#156)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| LMArena Longer Query | 1301 | 1327 |
Writing & Preference Mistral Small leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), Mistral Small: 52.5 (#171)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Mistral Small |
|---|---|---|
| LMArena Text | 1294 | 1338 |
| LMArena Creative Writing | 1285 | 1305 |
| LMArena Multi-Turn | 1297 | 1344 |
| LiveBench Language | — | 30.5% |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than Mistral Small?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 33.4 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or Mistral Small better for coding?
Mistral Small scores higher on coding benchmarks: 34.0 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Mistral Small share?
19 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Mistral Small has 39.