Model comparison
GPT-5 vs Llama 3-8B
GPT-5 is the stronger model overall, scoring 50.9 to 25.3 on the Noometry Index. Llama 3-8B costs 25× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Last verified . 24 shared benchmarks.
MetaSummary
- They share 24 benchmarks with published results for both. GPT-5 scores higher in 8 categories and Llama 3-8B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-5 leads 56.5 to 7.8.
- The biggest single-benchmark swing is MATH Level 5: 98.1% for GPT-5 and 6.1% for Llama 3-8B.
- Llama 3-8B is cheaper at $0.14 / $0.14 per million input/output tokens, against $1.25 / $10 for GPT-5.
- GPT-5 accepts more context: 400K tokens versus 8K.
- Llama 3-8B has downloadable open weights; the other is API-only.
Side by side
| GPT-5 | Llama 3-8B | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 50.9 | 25.3 |
| Released | 2025-08-07 | 2024-04-18 |
| Weights | Proprietary | Open |
| Context window | 400K | 8K |
| Max output | 128K | 8K |
| Input $ / M tokens | $1.25 | $0.14 |
| Output $ / M tokens | $10 | $0.14 |
| Results tracked | 69 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
GPT-5: 50.7 (#39), Llama 3-8B: 30.5 (#285)
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| LMArena Coding | 1436 | 1152 |
| SWE-bench Verified | 73.6% | — |
| SWE-bench Verified (bash only) | 65% | — |
| Aider Polyglot | 88% | — |
| LMArena WebDev | 1417 | — |
| SciCode | 42.9% | — |
| GSO | 6.9% | — |
| WeirdML | 60.7% | — |
| BigCodeBench Instruct | — | 31.9% |
| BigCodeBench Complete | — | 36.9% |
| ALE-Bench | 1,162 | — |
| AlgoTune | 1.67 | — |
| HumanEval+ | — | 56.7% |
| MBPP+ | — | 54.8% |
Agentic & Tool Use Not comparable
GPT-5: 33.0 (#53), Llama 3-8B: —
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| Terminal-Bench | 49.6% | — |
| GDPval | 34.8% | — |
| Remote Labor Index | 1.7% | — |
| DeepResearch Bench | 49.6% | — |
| BALROG | 32.8% | — |
| LMArena Search | 1133 | — |
| METR Time Horizons | 69.6% | — |
Reasoning GPT-5 leads
GPT-5: 38.3 (#55), Llama 3-8B: 13.8 (#321)
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| Chess Puzzles | 37% | 0% |
| LMArena Hard Prompts | 1415 | 1133 |
| DTBench | 90.7% | 43.9% |
| Epoch Capabilities Index | 150 | 116.33 |
| ForecastBench | 61.4 | 58.6 |
| ARC-AGI-2 | 9.9% | — |
| SimpleBench | 56.7% | — |
| Kagi LLM Benchmark | 72.7% | — |
| ARC-AGI-1 | 65.7% | — |
| CritPt | 12.6% | — |
| EnigmaEval | 10.5% | — |
| EBR-Bench | 12.7% | — |
| Mystery Game Puzzles | 23% | — |
| LMCA | 40% | — |
| Adversarial NLI | — | 57.3% |
| WinoGrande | — | 75.7% |
Math GPT-5 leads
GPT-5: 54.6 (#36), Llama 3-8B: 8.3 (#317)
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 91.4% | 1.9% |
| LMArena Math | 1406 | 1151 |
| MATH Level 5 | 98.1% | 6.1% |
| FrontierMath (Tiers 1-3) | 55.4% | — |
| FrontierMath Tier 4 | 22% | — |
| ProofBench | 18% | — |
| Omni-MATH | 64.7% | — |
| FrontierMath (Feb 2025 set) | 32.4% | — |
| FrontierMath Tier 4 (v1) | 12.5% | — |
Knowledge GPT-5 leads
GPT-5: 56.5 (#37), Llama 3-8B: 7.8 (#303)
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| GPQA Diamond | 86.2% | 26.1% |
| LMArena Expert | 1420 | 1113 |
| Humanity's Last Exam | 25.3% | — |
| SimpleQA Verified | 50.1% | — |
| MMLU-Pro | 86.3% | — |
| Confabulations | 10.3% | — |
| Vectara Hallucination Rate | 14.7% | — |
| GPQA (HELM) | 79.2% | — |
| ARC (AI2) Challenge | — | 82.8% |
| MMLU | — | 68.8% |
| OpenBookQA | — | 82.6% |
| TriviaQA | — | 67.7% |
Multimodal Not comparable
GPT-5: 46.1 (#12), Llama 3-8B: —
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| LMArena Vision | 1232 | — |
| GeoBench | 81% | — |
| VPCT | 66% | — |
Multilingual GPT-5 leads
GPT-5: 51.5 (#105), Llama 3-8B: 31.0 (#256)
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| LMArena Non-English | 1397 | 1098 |
| LMArena Chinese | 1423 | 1076 |
| LMArena French | 1409 | 1158 |
| LMArena German | 1415 | 1104 |
| LMArena Japanese | 1411 | 967 |
| LMArena Korean | 1361 | 1003 |
| LMArena Russian | 1406 | 1109 |
| LMArena Spanish | 1398 | 1172 |
Instruction Following GPT-5 leads
GPT-5: 74.0 (#107), Llama 3-8B: 58.7 (#255)
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| LMArena Instruction Following | 1388 | 1127 |
| IFEval | 87.5% | — |
Long Context GPT-5 leads
GPT-5: 69.5 (#2), Llama 3-8B: 34.2 (#246)
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| LMArena Longer Query | 1399 | 1128 |
| Fiction.LiveBench | 97.2% | — |
Writing & Preference GPT-5 leads
GPT-5: 63.5 (#63), Llama 3-8B: 37.7 (#251)
| Benchmark | GPT-5 | Llama 3-8B |
|---|---|---|
| LMArena Text | 1406 | 1166 |
| LMArena Creative Writing | 1365 | 1150 |
| LMArena Multi-Turn | 1426 | 1152 |
| Short-Story Creative Writing | 86% | — |
| EQ-Bench Creative Writing | 1627 | — |
| WildBench | 85.7% | — |
Frequently asked questions
Is GPT-5 better than Llama 3-8B?
GPT-5 is the stronger model overall, scoring 50.9 to 25.3 on the Noometry Index. Llama 3-8B costs 25× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Which is cheaper, GPT-5 or Llama 3-8B?
Llama 3-8B is cheaper. It lists at $0.14 per million input tokens and $0.14 per million output tokens; GPT-5 lists at $1.25 and $10.
Is GPT-5 or Llama 3-8B better for coding?
GPT-5 scores higher on coding benchmarks: 50.7 versus 30.5 in the Noometry coding category.
Which has the bigger context window?
GPT-5 does, with 400K tokens against 8K.
How many benchmarks do GPT-5 and Llama 3-8B share?
24 benchmarks have published results for both models. GPT-5 has 69 scored results on Noometry and Llama 3-8B has 34.