Model comparison
DeepSeek-V3.2-Exp vs Llama 2-70B
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 24.4 on the Noometry Index.
Last verified . 21 shared benchmarks.
DeepSeek
MetaSummary
- They share 21 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 8 categories and Llama 2-70B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3.2-Exp leads 51.7 to 7.4.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 87.8% for DeepSeek-V3.2-Exp and 0% for Llama 2-70B.
Side by side
| DeepSeek-V3.2-Exp | Llama 2-70B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 44.3 | 24.4 |
| Released | 2025-09-29 | 2023-07-18 |
| Weights | Open | Open |
| Context window | 164K | — |
| Max output | 66K | — |
| Input $ / M tokens | $0.26 | — |
| Output $ / M tokens | $0.38 | — |
| Results tracked | 49 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 46.5 (#65), Llama 2-70B: 31.4 (#286)
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| LMArena Coding | 1454 | 1079 |
| 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-V3.2-Exp: 32.7 (#59), Llama 2-70B: —
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| Terminal-Bench | 39.6% | — |
| APEX-Agents | 21.3% | — |
| Berkeley Function Calling Leaderboard | 56.7% | — |
| TheAgentCompany | 42.9% | — |
| Vending-Bench 2 | 1,034 | — |
Reasoning DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 22.1 (#208), Llama 2-70B: 14.4 (#325)
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| LMArena Hard Prompts | 1434 | 1073 |
| DTBench | 87.7% | 41.6% |
| Epoch Capabilities Index | 146.27 | 113.79 |
| ARC-AGI-2 | 4% | — |
| Kagi LLM Benchmark | 52.2% | — |
| NYT Connections (extended) | 36.7% | — |
| ARC-AGI-1 | 57% | — |
| CritPt | 2.9% | — |
| Chess Puzzles | 14% | — |
| Thematic Generalization | 65% | — |
| LMCA | 29.1% | — |
| BIG-Bench Hard | — | 64.9% |
| CommonsenseQA 2.0 | — | 50% |
| ForecastBench | — | 51.4 |
| HellaSwag | — | 85.3% |
| LAMBADA | — | 78.9% |
| PIQA | — | 82.8% |
| WinoGrande | — | 80.2% |
Math DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 41.7 (#87), Llama 2-70B: 8.1 (#326)
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 87.8% | 0% |
| LMArena Math | 1435 | 1091 |
| MathArena Final-Answer Competitions | 57.7% | — |
| ProofBench | 8% | — |
| MATH Level 5 | — | 3.3% |
| FrontierMath (Feb 2025 set) | 22.1% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
| GSM8K | — | 69.6% |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 51.7 (#66), Llama 2-70B: 7.4 (#310)
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| GPQA Diamond | 83.4% | 26.3% |
| LMArena Expert | 1436 | 1039 |
| Vectara Hallucination Rate | 5.3% | — |
| ARC (AI2) Challenge | — | 78.3% |
| BoolQ | — | 88.6% |
| MMLU | — | 69.9% |
| OpenBookQA | — | 60.2% |
| TriviaQA | — | 87.6% |
Multilingual DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 52.2 (#90), Llama 2-70B: 27.7 (#274)
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| LMArena Non-English | 1409 | 1045 |
| LMArena Chinese | 1461 | 995 |
| LMArena French | 1433 | 1090 |
| LMArena German | 1440 | 1041 |
| LMArena Japanese | 1374 | 927 |
| LMArena Korean | 1371 | 964 |
| LMArena Russian | 1424 | 1083 |
| LMArena Spanish | 1440 | 1143 |
Instruction Following DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 74.5 (#93), Llama 2-70B: 54.9 (#278)
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| LMArena Instruction Following | 1413 | 1071 |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#16), Llama 2-70B: 32.3 (#270)
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| LMArena Longer Query | 1428 | 1062 |
| Fiction.LiveBench | 83.3% | — |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 62.4 (#77), Llama 2-70B: 32.3 (#279)
| Benchmark | DeepSeek-V3.2-Exp | Llama 2-70B |
|---|---|---|
| LMArena Text | 1425 | 1115 |
| LMArena Creative Writing | 1403 | 1075 |
| LMArena Multi-Turn | 1427 | 1088 |
| EQ-Bench Creative Writing | 1515 | — |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than Llama 2-70B?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 24.4 on the Noometry Index.
Is DeepSeek-V3.2-Exp or Llama 2-70B better for coding?
DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.5 versus 31.4 in the Noometry coding category.
How many benchmarks do DeepSeek-V3.2-Exp and Llama 2-70B share?
21 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 49 scored results on Noometry and Llama 2-70B has 35.