Model comparison
DeepSeek-V2.5 (Sep 2024) vs DeepSeek-V3.1-Terminus
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 43.1 to 37.6 on the Noometry Index.
Last verified . 10 shared benchmarks.
DeepSeekSummary
- They share 10 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 0 categories and DeepSeek-V3.1-Terminus in 7 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3.1-Terminus leads 61.0 to 49.8.
Side by side
| DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus | |
|---|---|---|
| Provider | DeepSeek | DeepSeek |
| Noometry Index | 37.6 | 43.1 |
| Released | 2024-09-06 | 2025-09-22 |
| Weights | Open | Open |
| Context window | — | 164K |
| Max output | — | 147K |
| Input $ / M tokens | — | $0.27 |
| Output $ / M tokens | — | $1 |
| Results tracked | 22 | 16 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.1-Terminus leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), DeepSeek-V3.1-Terminus: 42.0 (#113)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Coding | 1309 | 1426 |
| Aider Polyglot | 17.8% | — |
| SciCode | — | 40.6% |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| ALE-Bench | — | 745.17 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Reasoning Too close to call
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), DeepSeek-V3.1-Terminus: 26.4 (#133)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1426 |
| Kagi LLM Benchmark | — | 57.4% |
| CritPt | — | 1.7% |
| DTBench | — | 81.3% |
| LMCA | — | 28.6% |
Math DeepSeek-V3.1-Terminus leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), DeepSeek-V3.1-Terminus: 38.5 (#137)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Math | 1288 | 1402 |
Knowledge Not comparable
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), DeepSeek-V3.1-Terminus: —
| Benchmark | DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Expert | 1266 | — |
Multilingual DeepSeek-V3.1-Terminus leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), DeepSeek-V3.1-Terminus: 52.1 (#92)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Non-English | 1273 | 1407 |
| LMArena Russian | 1289 | 1436 |
| LMArena Chinese | 1318 | — |
| LMArena French | 1289 | — |
| LMArena German | 1258 | — |
| LMArena Japanese | 1228 | — |
| LMArena Korean | 1209 | — |
| LMArena Spanish | 1248 | — |
Instruction Following DeepSeek-V3.1-Terminus leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), DeepSeek-V3.1-Terminus: 74.0 (#106)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Instruction Following | 1280 | 1404 |
Long Context DeepSeek-V3.1-Terminus leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), DeepSeek-V3.1-Terminus: 43.4 (#97)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Longer Query | 1301 | 1421 |
Writing & Preference DeepSeek-V3.1-Terminus leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), DeepSeek-V3.1-Terminus: 61.0 (#92)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Text | 1294 | 1419 |
| LMArena Creative Writing | 1285 | 1403 |
| LMArena Multi-Turn | 1297 | 1411 |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than DeepSeek-V3.1-Terminus?
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 43.1 to 37.6 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or DeepSeek-V3.1-Terminus better for coding?
DeepSeek-V3.1-Terminus scores higher on coding benchmarks: 42.0 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and DeepSeek-V3.1-Terminus share?
10 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and DeepSeek-V3.1-Terminus has 16.