Model comparison
DeepSeek-V3.1-Terminus vs Step 3
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 43.1 to 40.5 on the Noometry Index.
Last verified . 11 shared benchmarks.
DeepSeek
StepFunSummary
- They share 11 benchmarks with published results for both. DeepSeek-V3.1-Terminus scores higher in 6 categories and Step 3 in 1 category; 6 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3.1-Terminus leads 61.0 to 54.3.
Side by side
| DeepSeek-V3.1-Terminus | Step 3 | |
|---|---|---|
| Provider | DeepSeek | StepFun |
| Noometry Index | 43.1 | 40.5 |
| Released | 2025-09-22 | — |
| Weights | Open | Open |
| Context window | 164K | — |
| Max output | 147K | — |
| Input $ / M tokens | $0.27 | — |
| Output $ / M tokens | $1 | — |
| Results tracked | 16 | 17 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 42.0 (#113), Step 3: 40.1 (#147)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| LMArena Coding | 1426 | 1367 |
| SciCode | 40.6% | — |
| ALE-Bench | 745.17 | — |
Reasoning Step 3 leads
DeepSeek-V3.1-Terminus: 26.4 (#133), Step 3: 28.4 (#105)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| Kagi LLM Benchmark | 57.4% | 62.3% |
| LMArena Hard Prompts | 1426 | 1355 |
| CritPt | 1.7% | — |
| DTBench | 81.3% | — |
| LMCA | 28.6% | — |
Math Too close to call
DeepSeek-V3.1-Terminus: 38.5 (#137), Step 3: 37.6 (#148)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| LMArena Math | 1402 | 1366 |
Knowledge Not comparable
DeepSeek-V3.1-Terminus: —, Step 3: 36.8 (#164)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| LMArena Expert | — | 1333 |
Multimodal Not comparable
DeepSeek-V3.1-Terminus: —, Step 3: 35.5 (#86)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| LMArena Vision | — | 1177 |
Multilingual DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 52.1 (#92), Step 3: 46.3 (#159)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| LMArena Non-English | 1407 | 1327 |
| LMArena Russian | 1436 | 1331 |
| LMArena Chinese | — | 1397 |
| LMArena German | — | 1371 |
| LMArena Korean | — | 1269 |
| LMArena Spanish | — | 1371 |
Instruction Following DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 74.0 (#106), Step 3: 70.4 (#164)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| LMArena Instruction Following | 1404 | 1332 |
Long Context DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 43.4 (#97), Step 3: 40.3 (#157)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| LMArena Longer Query | 1421 | 1326 |
Writing & Preference DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 61.0 (#92), Step 3: 54.3 (#151)
| Benchmark | DeepSeek-V3.1-Terminus | Step 3 |
|---|---|---|
| LMArena Text | 1419 | 1350 |
| LMArena Creative Writing | 1403 | 1321 |
| LMArena Multi-Turn | 1411 | 1341 |
Frequently asked questions
Is DeepSeek-V3.1-Terminus better than Step 3?
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 43.1 to 40.5 on the Noometry Index.
Is DeepSeek-V3.1-Terminus or Step 3 better for coding?
DeepSeek-V3.1-Terminus scores higher on coding benchmarks: 42.0 versus 40.1 in the Noometry coding category.
How many benchmarks do DeepSeek-V3.1-Terminus and Step 3 share?
11 benchmarks have published results for both models. DeepSeek-V3.1-Terminus has 16 scored results on Noometry and Step 3 has 17.