Model comparison
GPT-4o mini vs Llama 3.2 3B
Llama 3.2 3B is the stronger model overall, scoring 28.9 to 25.5 on the Noometry Index.
Last verified . 17 shared benchmarks.
MetaSummary
- They share 17 benchmarks with published results for both. GPT-4o mini scores higher in 5 categories and Llama 3.2 3B in 4 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Llama 3.2 3B leads 32.4 to 10.4.
- The biggest single-benchmark swing is BigCodeBench Complete: 57.4% for GPT-4o mini and 28.3% for Llama 3.2 3B.
- Llama 3.2 3B is cheaper at $0.05 / $0.33 per million input/output tokens, against $0.15 / $0.60 for GPT-4o mini.
- Llama 3.2 3B accepts more context: 131K tokens versus 128K.
- Llama 3.2 3B has downloadable open weights; the other is API-only.
Side by side
| GPT-4o mini | Llama 3.2 3B | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 25.5 | 28.9 |
| Released | 2024-07-18 | 2024-09-24 |
| Weights | Proprietary | Open |
| Context window | 128K | 131K |
| Max output | 16K | 118K |
| Input $ / M tokens | $0.15 | $0.05 |
| Output $ / M tokens | $0.60 | $0.33 |
| Results tracked | 60 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 3.2 3B leads
GPT-4o mini: 22.0 (#335), Llama 3.2 3B: 27.6 (#319)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| BigCodeBench Instruct | 46.1% | 23.4% |
| LMArena Coding | 1290 | 1098 |
| BigCodeBench Complete | 57.4% | 28.3% |
| Aider Polyglot | 3.6% | — |
| WeirdML | 11.8% | — |
| LiveBench Coding | 43.1% | — |
| HumanEval+ | 83.5% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use GPT-4o mini leads
GPT-4o mini: 27.5 (#101), Llama 3.2 3B: 20.1 (#143)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| BALROG | 17.4% | 10.1% |
| Berkeley Function Calling Leaderboard | — | 21.9% |
Reasoning Llama 3.2 3B leads
GPT-4o mini: 8.7 (#347), Llama 3.2 3B: 21.0 (#228)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| LMArena Hard Prompts | 1267 | 1095 |
| ARC-AGI-2 | 0% | — |
| SimpleBench | 10.7% | — |
| Kagi LLM Benchmark | 28.8% | — |
| Chess Puzzles | 0% | — |
| LiveBench Reasoning | 32.8% | — |
| Mystery Game Puzzles | 12% | — |
| DTBench | 54.4% | — |
| LiveBench Data Analysis | 50% | — |
| LMCA | 10.4% | — |
| Epoch Capabilities Index | 126.56 | — |
| LiveBench | 41.3% | — |
| PIQA | 88.7% | — |
Math Llama 3.2 3B leads
GPT-4o mini: 10.4 (#314), Llama 3.2 3B: 32.4 (#214)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| LMArena Math | 1267 | 1126 |
| FrontierMath (Tiers 1-3) | 0.7% | — |
| OTIS Mock AIME 2024-2025 | 6.9% | — |
| Omni-MATH | 28% | — |
| LiveBench Math | 36.3% | — |
| MATH Level 5 | 52.6% | — |
| GSM8K | 91.3% | — |
Knowledge Llama 3.2 3B leads
GPT-4o mini: 17.7 (#284), Llama 3.2 3B: 29.7 (#235)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| LMArena Expert | 1235 | 1090 |
| GPQA Diamond | 37.7% | — |
| SimpleQA Verified | 8.3% | — |
| MMLU-Pro | 60.3% | — |
| Confabulations | 37.2% | — |
| GPQA (HELM) | 36.8% | — |
| BoolQ | 88.7% | — |
| MMLU | 81.8% | — |
Multimodal Not comparable
GPT-4o mini: 25.9 (#122), Llama 3.2 3B: —
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| LMArena Vision | 1066 | — |
| Video-MME | 64.8% | — |
| GeoBench | 64% | — |
| VPCT | 34% | — |
Multilingual GPT-4o mini leads
GPT-4o mini: 42.0 (#199), Llama 3.2 3B: 26.2 (#281)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| LMArena Non-English | 1266 | 1019 |
| LMArena Chinese | 1265 | 1017 |
| LMArena German | 1272 | 1056 |
| LMArena Russian | 1275 | 949 |
| LMArena French | 1297 | — |
| LMArena Japanese | 1216 | — |
| LMArena Korean | 1195 | — |
| LMArena Spanish | 1276 | — |
Instruction Following GPT-4o mini leads
GPT-4o mini: 61.9 (#239), Llama 3.2 3B: 56.0 (#275)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| LMArena Instruction Following | 1258 | 1089 |
| LiveBench Instruction Following | 56.8% | — |
| IFEval | 78.2% | — |
Long Context GPT-4o mini leads
GPT-4o mini: 39.1 (#186), Llama 3.2 3B: 33.4 (#261)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| LMArena Longer Query | 1289 | 1100 |
Writing & Preference GPT-4o mini leads
GPT-4o mini: 39.5 (#248), Llama 3.2 3B: 24.7 (#307)
| Benchmark | GPT-4o mini | Llama 3.2 3B |
|---|---|---|
| LMArena Text | 1286 | 1110 |
| LMArena Creative Writing | 1268 | 1094 |
| EQ-Bench Creative Writing | 873 | 595 |
| LMArena Multi-Turn | 1285 | 1105 |
| Short-Story Creative Writing | 67.2% | — |
| WildBench | 79.1% | — |
| LiveBench Language | 28.6% | — |
Frequently asked questions
Is GPT-4o mini better than Llama 3.2 3B?
Llama 3.2 3B is the stronger model overall, scoring 28.9 to 25.5 on the Noometry Index.
Which is cheaper, GPT-4o mini or Llama 3.2 3B?
Llama 3.2 3B is cheaper. It lists at $0.05 per million input tokens and $0.33 per million output tokens; GPT-4o mini lists at $0.15 and $0.60.
Is GPT-4o mini or Llama 3.2 3B better for coding?
Llama 3.2 3B scores higher on coding benchmarks: 27.6 versus 22.0 in the Noometry coding category.
Which has the bigger context window?
Llama 3.2 3B does, with 131K tokens against 128K.
How many benchmarks do GPT-4o mini and Llama 3.2 3B share?
17 benchmarks have published results for both models. GPT-4o mini has 60 scored results on Noometry and Llama 3.2 3B has 18.