Model comparison
DeepSeek-V3.1-Terminus vs Llama 3.1-70B
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 43.1 to 29.6 on the Noometry Index.
Last verified . 12 shared benchmarks.
DeepSeek
MetaSummary
- They share 12 benchmarks with published results for both. DeepSeek-V3.1-Terminus scores higher in 7 categories and Llama 3.1-70B in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3.1-Terminus leads 61.0 to 35.4.
- The biggest single-benchmark swing is DTBench: 81.3% for DeepSeek-V3.1-Terminus and 60% for Llama 3.1-70B.
- Llama 3.1-70B is cheaper at $0.40 / $0.40 per million input/output tokens, against $0.27 / $1 for DeepSeek-V3.1-Terminus.
- DeepSeek-V3.1-Terminus accepts more context: 164K tokens versus 128K.
Side by side
| DeepSeek-V3.1-Terminus | Llama 3.1-70B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 43.1 | 29.6 |
| Released | 2025-09-22 | 2024-07-23 |
| Weights | Open | Open |
| Context window | 164K | 128K |
| Max output | 147K | 4K |
| Input $ / M tokens | $0.27 | $0.40 |
| Output $ / M tokens | $1 | $0.40 |
| Results tracked | 16 | 35 |
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), Llama 3.1-70B: 30.3 (#296)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| LMArena Coding | 1426 | 1260 |
| SciCode | 40.6% | — |
| WeirdML | — | 9% |
| BigCodeBench Instruct | — | 46.1% |
| BigCodeBench Complete | — | 54.8% |
| ALE-Bench | 745.17 | — |
Agentic & Tool Use Not comparable
DeepSeek-V3.1-Terminus: —, Llama 3.1-70B: 25.1 (#112)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| TheAgentCompany | — | 6.9% |
| BALROG | — | 27.9% |
Reasoning DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 26.4 (#133), Llama 3.1-70B: 21.6 (#220)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1426 | 1241 |
| DTBench | 81.3% | 60% |
| LMCA | 28.6% | 14.8% |
| Kagi LLM Benchmark | 57.4% | — |
| CritPt | 1.7% | — |
| Epoch Capabilities Index | — | 125.92 |
Math DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 38.5 (#137), Llama 3.1-70B: 13.5 (#304)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| LMArena Math | 1402 | 1252 |
| OTIS Mock AIME 2024-2025 | — | 3.6% |
| Omni-MATH | — | 21% |
| MATH Level 5 | — | 36.7% |
Knowledge Not comparable
DeepSeek-V3.1-Terminus: —, Llama 3.1-70B: 24.2 (#269)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| GPQA Diamond | — | 44.2% |
| MMLU-Pro | — | 65.3% |
| GPQA (HELM) | — | 42.6% |
| LMArena Expert | — | 1209 |
| MMLU | — | 80.1% |
Multilingual DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 52.1 (#92), Llama 3.1-70B: 38.8 (#225)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 1407 | 1219 |
| LMArena Russian | 1436 | 1234 |
| LMArena Chinese | — | 1215 |
| LMArena French | — | 1261 |
| LMArena German | — | 1222 |
| LMArena Japanese | — | 1132 |
| LMArena Korean | — | 1140 |
| LMArena Spanish | — | 1253 |
Instruction Following DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 74.0 (#106), Llama 3.1-70B: 65.3 (#223)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| LMArena Instruction Following | 1404 | 1231 |
| IFEval | — | 82.1% |
Long Context DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 43.4 (#97), Llama 3.1-70B: 37.6 (#214)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1421 | 1241 |
Writing & Preference DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 61.0 (#92), Llama 3.1-70B: 35.4 (#267)
| Benchmark | DeepSeek-V3.1-Terminus | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1419 | 1261 |
| LMArena Creative Writing | 1403 | 1232 |
| LMArena Multi-Turn | 1411 | 1256 |
| EQ-Bench Creative Writing | — | 784 |
| WildBench | — | 75.8% |
Frequently asked questions
Is DeepSeek-V3.1-Terminus better than Llama 3.1-70B?
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 43.1 to 29.6 on the Noometry Index.
Which is cheaper, DeepSeek-V3.1-Terminus or Llama 3.1-70B?
Llama 3.1-70B is cheaper. It lists at $0.40 per million input tokens and $0.40 per million output tokens; DeepSeek-V3.1-Terminus lists at $0.27 and $1.
Is DeepSeek-V3.1-Terminus or Llama 3.1-70B better for coding?
DeepSeek-V3.1-Terminus scores higher on coding benchmarks: 42.0 versus 30.3 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3.1-Terminus does, with 164K tokens against 128K.
How many benchmarks do DeepSeek-V3.1-Terminus and Llama 3.1-70B share?
12 benchmarks have published results for both models. DeepSeek-V3.1-Terminus has 16 scored results on Noometry and Llama 3.1-70B has 35.