Model comparison
Llama 3.1-70B vs o1
o1 is the stronger model overall, scoring 40.9 to 29.6 on the Noometry Index. Llama 3.1-70B costs 66× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Last verified . 24 shared benchmarks.
MetaSummary
- They share 24 benchmarks with published results for both. Llama 3.1-70B scores higher in 1 category and o1 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where o1 leads 36.1 to 13.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 3.6% for Llama 3.1-70B and 73.3% for o1.
- Llama 3.1-70B is cheaper at $0.40 / $0.40 per million input/output tokens, against $15 / $60 for o1.
- o1 accepts more context: 200K tokens versus 128K.
- Llama 3.1-70B has downloadable open weights; the other is API-only.
Side by side
| Llama 3.1-70B | o1 | |
|---|---|---|
| Provider | Meta | OpenAI |
| Noometry Index | 29.6 | 40.9 |
| Released | 2024-07-23 | 2024-09-12 |
| Weights | Open | Proprietary |
| Context window | 128K | 200K |
| Max output | 4K | 100K |
| Input $ / M tokens | $0.40 | $15 |
| Output $ / M tokens | $0.40 | $60 |
| Results tracked | 35 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1 leads
Llama 3.1-70B: 30.3 (#296), o1: 46.1 (#70)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| WeirdML | 9% | 47.6% |
| LMArena Coding | 1260 | 1367 |
| Aider Polyglot | — | 61.7% |
| BigCodeBench Instruct | 46.1% | — |
| LiveBench Coding | — | 69.7% |
| BigCodeBench Complete | 54.8% | — |
| CadEval | — | 56% |
| HumanEval+ | — | 89% |
| MBPP+ | — | 80.2% |
Agentic & Tool Use Too close to call
Llama 3.1-70B: 25.1 (#112), o1: 24.6 (#117)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| TheAgentCompany | 6.9% | — |
| Cybench | — | 10% |
| BALROG | 27.9% | — |
| METR Time Horizons | — | 51.1% |
Reasoning o1 leads
Llama 3.1-70B: 21.6 (#220), o1: 27.9 (#111)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1371 |
| DTBench | 60% | 74.7% |
| LMCA | 14.8% | 22.3% |
| Epoch Capabilities Index | 125.92 | 141.91 |
| SimpleBench | — | 41.7% |
| ARC-AGI-1 | — | 30.7% |
| Chess Puzzles | — | 15% |
| EnigmaEval | — | 5.7% |
| LiveBench Reasoning | — | 91.6% |
| LiveBench Data Analysis | — | 65.5% |
| LiveBench | — | 75.7% |
Math o1 leads
Llama 3.1-70B: 13.5 (#304), o1: 36.1 (#175)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 3.6% | 73.3% |
| LMArena Math | 1252 | 1388 |
| MATH Level 5 | 36.7% | 94.7% |
| FrontierMath (Tiers 1-3) | — | 14.7% |
| Omni-MATH | 21% | — |
| LiveBench Math | — | 80.3% |
| FrontierMath (Feb 2025 set) | — | 9.3% |
Knowledge o1 leads
Llama 3.1-70B: 24.2 (#269), o1: 41.5 (#110)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| GPQA Diamond | 44.2% | 76.8% |
| LMArena Expert | 1209 | 1361 |
| Humanity's Last Exam | — | 8% |
| SimpleQA Verified | — | 41.1% |
| MMLU-Pro | 65.3% | — |
| Confabulations | — | 11.7% |
| GPQA (HELM) | 42.6% | — |
| MMLU | 80.1% | — |
Multimodal Not comparable
Llama 3.1-70B: —, o1: 34.2 (#93)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| LMArena Vision | — | 1168 |
| GeoBench | — | 80% |
| VPCT | — | 37% |
| SpatialViz-Bench | — | 41.4% |
Multilingual o1 leads
Llama 3.1-70B: 38.8 (#225), o1: 48.6 (#142)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| LMArena Non-English | 1219 | 1358 |
| LMArena Chinese | 1215 | 1394 |
| LMArena French | 1261 | 1344 |
| LMArena German | 1222 | 1337 |
| LMArena Japanese | 1132 | 1346 |
| LMArena Korean | 1140 | 1396 |
| LMArena Russian | 1234 | 1356 |
| LMArena Spanish | 1253 | 1345 |
Instruction Following o1 leads
Llama 3.1-70B: 65.3 (#223), o1: 74.8 (#86)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| LMArena Instruction Following | 1231 | 1367 |
| LiveBench Instruction Following | — | 81.5% |
| IFEval | 82.1% | — |
Long Context o1 leads
Llama 3.1-70B: 37.6 (#214), o1: 50.3 (#9)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| LMArena Longer Query | 1241 | 1378 |
| Fiction.LiveBench | — | 83.3% |
Writing & Preference o1 leads
Llama 3.1-70B: 35.4 (#267), o1: 55.6 (#144)
| Benchmark | Llama 3.1-70B | o1 |
|---|---|---|
| LMArena Text | 1261 | 1366 |
| LMArena Creative Writing | 1232 | 1348 |
| LMArena Multi-Turn | 1256 | 1369 |
| Short-Story Creative Writing | — | 70.2% |
| EQ-Bench Creative Writing | 784 | — |
| WildBench | 75.8% | — |
| LiveBench Language | — | 65.4% |
Frequently asked questions
Is Llama 3.1-70B better than o1?
o1 is the stronger model overall, scoring 40.9 to 29.6 on the Noometry Index. Llama 3.1-70B costs 66× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Which is cheaper, Llama 3.1-70B or o1?
Llama 3.1-70B is cheaper. It lists at $0.40 per million input tokens and $0.40 per million output tokens; o1 lists at $15 and $60.
Is Llama 3.1-70B or o1 better for coding?
o1 scores higher on coding benchmarks: 46.1 versus 30.3 in the Noometry coding category.
Which has the bigger context window?
o1 does, with 200K tokens against 128K.
How many benchmarks do Llama 3.1-70B and o1 share?
24 benchmarks have published results for both models. Llama 3.1-70B has 35 scored results on Noometry and o1 has 52.