Model comparison
DeepSeek LLM 67B vs DeepSeek-V3.2-Exp
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.9 to 24.8 on the Noometry Index.
Last verified . 14 shared benchmarks.
DeepSeekSummary
- They share 14 benchmarks with published results for both. DeepSeek LLM 67B scores higher in 0 categories and DeepSeek-V3.2-Exp in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3.2-Exp leads 51.6 to 7.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 0.8% for DeepSeek LLM 67B and 87.8% for DeepSeek-V3.2-Exp.
Side by side
| DeepSeek LLM 67B | DeepSeek-V3.2-Exp | |
|---|---|---|
| Provider | DeepSeek | DeepSeek |
| Noometry Index | 24.8 | 44.9 |
| Released | 2023-11-29 | 2025-09-29 |
| Weights | Open | Open |
| Context window | — | 164K |
| Max output | — | 66K |
| Input $ / M tokens | — | $0.26 |
| Output $ / M tokens | — | $0.38 |
| Results tracked | 15 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.2-Exp leads
DeepSeek LLM 67B: 31.3 (#273), DeepSeek-V3.2-Exp: 46.4 (#61)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Coding | 1096 | 1454 |
| SWE-bench Verified (bash only) | — | 70% |
| Aider Polyglot | — | 74.2% |
| LMArena WebDev | — | 1362 |
| SWE-bench Multilingual | — | 59% |
| SciCode | — | 38.9% |
| WeirdML | — | 39.5% |
Agentic & Tool Use Not comparable
DeepSeek LLM 67B: —, DeepSeek-V3.2-Exp: 37.9 (#31)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| Terminal-Bench | — | 39.6% |
| Berkeley Function Calling Leaderboard | — | 56.7% |
| TheAgentCompany | — | 42.9% |
| Vending-Bench 2 | — | 1,034 |
Reasoning DeepSeek-V3.2-Exp leads
DeepSeek LLM 67B: 15.9 (#300), DeepSeek-V3.2-Exp: 22.0 (#185)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| Chess Puzzles | 0% | 14% |
| LMArena Hard Prompts | 1070 | 1433 |
| Epoch Capabilities Index | 110.54 | 146.28 |
| ARC-AGI-2 | — | 4% |
| Kagi LLM Benchmark | — | 52.2% |
| NYT Connections (extended) | — | 36.7% |
| ARC-AGI-1 | — | 57% |
| CritPt | — | 2.9% |
| Thematic Generalization | — | 65% |
| DTBench | — | 87.7% |
| LMCA | — | 29.1% |
Math DeepSeek-V3.2-Exp leads
DeepSeek LLM 67B: 8.1 (#318), DeepSeek-V3.2-Exp: 40.8 (#75)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 0.8% | 87.8% |
| LMArena Math | 1108 | 1435 |
| MathArena Final-Answer Competitions | — | 57.7% |
| ProofBench | — | 8% |
| MATH Level 5 | 6.4% | — |
| FrontierMath (Feb 2025 set) | — | 22.1% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek LLM 67B: 7.1 (#308), DeepSeek-V3.2-Exp: 51.6 (#61)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| GPQA Diamond | 24.6% | 83.4% |
| Vectara Hallucination Rate | — | 5.3% |
| LMArena Expert | — | 1436 |
Multilingual DeepSeek-V3.2-Exp leads
DeepSeek LLM 67B: 29.5 (#262), DeepSeek-V3.2-Exp: 52.4 (#85)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Non-English | 1073 | 1409 |
| LMArena Chinese | 1132 | 1461 |
| LMArena French | — | 1435 |
| LMArena German | — | 1440 |
| LMArena Japanese | — | 1377 |
| LMArena Korean | — | 1372 |
| LMArena Russian | — | 1423 |
| LMArena Spanish | — | 1441 |
Instruction Following DeepSeek-V3.2-Exp leads
DeepSeek LLM 67B: 55.7 (#272), DeepSeek-V3.2-Exp: 74.7 (#89)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Instruction Following | 1079 | 1413 |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek LLM 67B: 33.1 (#260), DeepSeek-V3.2-Exp: 47.6 (#15)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Longer Query | 1092 | 1428 |
| Fiction.LiveBench | — | 83.3% |
| CL-bench | — | 13.2% |
| CL-bench Life | — | 9.5% |
Writing & Preference DeepSeek-V3.2-Exp leads
DeepSeek LLM 67B: 31.8 (#277), DeepSeek-V3.2-Exp: 62.5 (#73)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Text | 1105 | 1425 |
| LMArena Creative Writing | 1067 | 1403 |
| LMArena Multi-Turn | 1082 | 1427 |
| EQ-Bench Creative Writing | — | 1515 |
Frequently asked questions
Is DeepSeek LLM 67B better than DeepSeek-V3.2-Exp?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.9 to 24.8 on the Noometry Index.
Is DeepSeek LLM 67B or DeepSeek-V3.2-Exp better for coding?
DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.4 versus 31.3 in the Noometry coding category.
How many benchmarks do DeepSeek LLM 67B and DeepSeek-V3.2-Exp share?
14 benchmarks have published results for both models. DeepSeek LLM 67B has 15 scored results on Noometry and DeepSeek-V3.2-Exp has 48.