Model comparison
GPT-5.2 vs Llama-3.3-70B-Instruct
GPT-5.2 is the stronger model overall, scoring 54.1 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 31× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.
Last verified . 27 shared benchmarks.
MetaSummary
- They share 27 benchmarks with published results for both. GPT-5.2 scores higher in 9 categories and Llama-3.3-70B-Instruct in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.2 leads 60.0 to 15.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 96.1% for GPT-5.2 and 5.1% for Llama-3.3-70B-Instruct.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $1.75 / $14 for GPT-5.2.
- GPT-5.2 accepts more context: 400K tokens versus 128K.
- Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.
Side by side
| GPT-5.2 | Llama-3.3-70B-Instruct | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 54.1 | 30.6 |
| Released | 2025-12-11 | 2024-12-06 |
| Weights | Proprietary | Open |
| Context window | 400K | 128K |
| Max output | 128K | 4K |
| Input $ / M tokens | $1.75 | $0.10 |
| Output $ / M tokens | $14 | $0.32 |
| Results tracked | 67 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.2 leads
GPT-5.2: 51.6 (#37), Llama-3.3-70B-Instruct: 31.0 (#290)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| WeirdML | 72.2% | 14.4% |
| LMArena Coding | 1447 | 1268 |
| SWE-bench Verified | 73.8% | — |
| SWE-bench Verified (bash only) | 72.8% | — |
| LMArena WebDev | 1416 | — |
| SWE-bench Multilingual | 66.7% | — |
| SciCode | — | 26% |
| GSO | 27.4% | — |
| BigCodeBench Instruct | — | 46.9% |
| LiveBench Coding | — | 36.6% |
| BigCodeBench Complete | — | 57.5% |
| ALE-Bench | 1,294 | — |
| AlgoTune | 2.05 | — |
Agentic & Tool Use GPT-5.2 leads
GPT-5.2: 40.2 (#24), Llama-3.3-70B-Instruct: 25.8 (#105)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| Berkeley Function Calling Leaderboard | 55.9% | 31.9% |
| Terminal-Bench | 64.9% | — |
| GDPval | 49.7% | — |
| Remote Labor Index | 2.5% | — |
| τ²-bench Airline | 83% | — |
| τ²-bench Banking | 32.2% | — |
| τ²-bench Retail | 81.6% | — |
| τ²-bench Telecom | 89.7% | — |
| DeepResearch Bench | 41.1% | — |
| BALROG | — | 23% |
| LMArena Search | 1207 | — |
| METR Time Horizons | 75.3% | — |
| Vending-Bench 2 | 3,591 | — |
Reasoning GPT-5.2 leads
GPT-5.2: 50.2 (#35), Llama-3.3-70B-Instruct: 14.1 (#327)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| SimpleBench | 45.8% | 19.9% |
| LMArena Hard Prompts | 1445 | 1257 |
| DTBench | 90.9% | 59.5% |
| LMCA | 43.9% | 17.5% |
| Epoch Capabilities Index | 153.45 | 127.33 |
| ForecastBench | 60.1 | 58.6 |
| ARC-AGI-2 | 52.9% | — |
| Kagi LLM Benchmark | 73.3% | — |
| NYT Connections (extended) | 83.6% | — |
| ARC-AGI-1 | 86.2% | — |
| CritPt | — | 0% |
| Chess Puzzles | 49% | — |
| EnigmaEval | 10.4% | — |
| EBR-Bench | 23% | — |
| LiveBench Reasoning | — | 50.8% |
| Mystery Game Puzzles | 23% | — |
| LiveBench Data Analysis | — | 49.5% |
| LiveBench | — | 50.2% |
Math GPT-5.2 leads
GPT-5.2: 60.0 (#38), Llama-3.3-70B-Instruct: 15.3 (#298)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 96.1% | 5.1% |
| LMArena Math | 1440 | 1267 |
| FrontierMath (Tiers 1-3) | 67.4% | — |
| FrontierMath Tier 4 | 31.7% | — |
| MathArena Final-Answer Competitions | 72% | — |
| ProofBench | 15% | — |
| LiveBench Math | — | 42.2% |
| MATH Level 5 | — | 41.6% |
| FrontierMath (Feb 2025 set) | 40.7% | — |
| FrontierMath Tier 4 (v1) | 18.8% | — |
Knowledge GPT-5.2 leads
GPT-5.2: 59.3 (#32), Llama-3.3-70B-Instruct: 30.6 (#226)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| GPQA Diamond | 91.4% | 47.4% |
| Vectara Hallucination Rate | 8.4% | 4.1% |
| LMArena Expert | 1445 | 1225 |
| Humanity's Last Exam | 27.8% | — |
| SimpleQA Verified | 37.1% | — |
| Confabulations | — | 22.8% |
| MMLU | — | 86.3% |
Multimodal Not comparable
GPT-5.2: 51.3 (#7), Llama-3.3-70B-Instruct: —
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Vision | 1268 | — |
| VPCT | 84% | — |
| Furniture Assembly | 38.3% | — |
| LMArena Document | 1405 | — |
Multilingual GPT-5.2 leads
GPT-5.2: 53.4 (#67), Llama-3.3-70B-Instruct: 39.9 (#220)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Non-English | 1425 | 1236 |
| LMArena Chinese | 1460 | 1217 |
| LMArena French | 1455 | 1281 |
| LMArena German | 1448 | 1251 |
| LMArena Japanese | 1420 | 1150 |
| LMArena Korean | 1392 | 1143 |
| LMArena Russian | 1440 | 1252 |
| LMArena Spanish | 1433 | 1270 |
Instruction Following GPT-5.2 leads
GPT-5.2: 74.7 (#89), Llama-3.3-70B-Instruct: 71.1 (#157)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Instruction Following | 1417 | 1242 |
| LiveBench Instruction Following | — | 82.7% |
Long Context GPT-5.2 leads
GPT-5.2: 44.0 (#78), Llama-3.3-70B-Instruct: 26.4 (#295)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Longer Query | 1428 | 1256 |
| Fiction.LiveBench | — | 33.3% |
| CL-bench | 18.2% | — |
Writing & Preference GPT-5.2 leads
GPT-5.2: 66.8 (#32), Llama-3.3-70B-Instruct: 47.6 (#207)
| Benchmark | GPT-5.2 | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Text | 1439 | 1274 |
| LMArena Creative Writing | 1401 | 1250 |
| LMArena Multi-Turn | 1458 | 1280 |
| EQ-Bench Creative Writing | 1703 | — |
| LiveBench Language | — | 39.2% |
Frequently asked questions
Is GPT-5.2 better than Llama-3.3-70B-Instruct?
GPT-5.2 is the stronger model overall, scoring 54.1 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 31× less per token, which makes it the better buy when GPT-5.2's lead doesn't matter for your workload.
Which is cheaper, GPT-5.2 or Llama-3.3-70B-Instruct?
Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; GPT-5.2 lists at $1.75 and $14.
Is GPT-5.2 or Llama-3.3-70B-Instruct better for coding?
GPT-5.2 scores higher on coding benchmarks: 51.6 versus 31.0 in the Noometry coding category.
Which has the bigger context window?
GPT-5.2 does, with 400K tokens against 128K.
How many benchmarks do GPT-5.2 and Llama-3.3-70B-Instruct share?
27 benchmarks have published results for both models. GPT-5.2 has 67 scored results on Noometry and Llama-3.3-70B-Instruct has 43.