Model comparison
Llama 3-8B vs Llama 4 Maverick
Llama 4 Maverick is the stronger model overall, scoring 30.6 to 25.3 on the Noometry Index. Llama 3-8B costs 2.2× less per token, which makes it the better buy when Llama 4 Maverick's lead doesn't matter for your workload.
Last verified . 25 shared benchmarks.
MetaSummary
- They share 25 benchmarks with published results for both. Llama 3-8B scores higher in 3 categories and Llama 4 Maverick in 5 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Llama 4 Maverick leads 33.3 to 7.8.
- The biggest single-benchmark swing is MATH Level 5: 6.1% for Llama 3-8B and 73% for Llama 4 Maverick.
- Llama 3-8B is cheaper at $0.14 / $0.14 per million input/output tokens, against $0.19 / $0.65 for Llama 4 Maverick.
- Llama 4 Maverick accepts more context: 128K tokens versus 8K.
Side by side
| Llama 3-8B | Llama 4 Maverick | |
|---|---|---|
| Provider | Meta | Meta |
| Noometry Index | 25.3 | 30.6 |
| Released | 2024-04-18 | 2025-04-05 |
| Weights | Open | Open |
| Context window | 8K | 128K |
| Max output | 8K | 4K |
| Input $ / M tokens | $0.14 | $0.19 |
| Output $ / M tokens | $0.14 | $0.65 |
| Results tracked | 34 | 54 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 3-8B leads
Llama 3-8B: 30.5 (#285), Llama 4 Maverick: 26.3 (#318)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| BigCodeBench Instruct | 31.9% | 49.7% |
| LMArena Coding | 1152 | 1302 |
| BigCodeBench Complete | 36.9% | 61.4% |
| SWE-bench Verified (bash only) | — | 21% |
| Aider Polyglot | — | 15.6% |
| SciCode | — | 33.1% |
| WeirdML | — | 24.5% |
| ALE-Bench | — | 172.97 |
| HumanEval+ | 56.7% | — |
| MBPP+ | 54.8% | — |
Agentic & Tool Use Not comparable
Llama 3-8B: —, Llama 4 Maverick: 27.9 (#86)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 37.3% |
Reasoning Llama 3-8B leads
Llama 3-8B: 13.8 (#321), Llama 4 Maverick: 10.1 (#337)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| LMArena Hard Prompts | 1133 | 1281 |
| DTBench | 43.9% | 61.9% |
| Epoch Capabilities Index | 116.33 | 132.2 |
| ForecastBench | 58.6 | 57.5 |
| ARC-AGI-2 | — | 0% |
| SimpleBench | — | 27.7% |
| Kagi LLM Benchmark | — | 55.9% |
| NYT Connections (extended) | — | 8% |
| ARC-AGI-1 | — | 4.4% |
| CritPt | — | 0% |
| Chess Puzzles | 0% | — |
| EnigmaEval | — | 0.6% |
| LMCA | — | 15.9% |
| Adversarial NLI | 57.3% | — |
| WinoGrande | 75.7% | — |
Math Llama 4 Maverick leads
Llama 3-8B: 8.3 (#317), Llama 4 Maverick: 24.6 (#256)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.9% | 20.6% |
| LMArena Math | 1151 | 1299 |
| MATH Level 5 | 6.1% | 73% |
| Omni-MATH | — | 42.2% |
| FrontierMath (Feb 2025 set) | — | 0.7% |
Knowledge Llama 4 Maverick leads
Llama 3-8B: 7.8 (#303), Llama 4 Maverick: 33.3 (#196)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| GPQA Diamond | 26.1% | 67% |
| LMArena Expert | 1113 | 1259 |
| Humanity's Last Exam | — | 5.7% |
| MMLU-Pro | — | 81% |
| Confabulations | — | 22.6% |
| Vectara Hallucination Rate | — | 8.2% |
| GPQA (HELM) | — | 65% |
| ARC (AI2) Challenge | 82.8% | — |
| MMLU | 68.8% | — |
| OpenBookQA | 82.6% | — |
| TriviaQA | 67.7% | — |
Multimodal Not comparable
Llama 3-8B: —, Llama 4 Maverick: 30.9 (#100)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| LMArena Vision | — | 1141 |
| GeoBench | — | 52% |
| SpatialViz-Bench | — | 31.8% |
Multilingual Llama 4 Maverick leads
Llama 3-8B: 31.0 (#256), Llama 4 Maverick: 42.4 (#190)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| LMArena Non-English | 1098 | 1269 |
| LMArena Chinese | 1076 | 1277 |
| LMArena French | 1158 | 1258 |
| LMArena German | 1104 | 1291 |
| LMArena Japanese | 967 | 1207 |
| LMArena Korean | 1003 | 1203 |
| LMArena Russian | 1109 | 1286 |
| LMArena Spanish | 1172 | 1293 |
Instruction Following Llama 4 Maverick leads
Llama 3-8B: 58.7 (#255), Llama 4 Maverick: 71.9 (#140)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| LMArena Instruction Following | 1127 | 1267 |
| IFEval | — | 90.8% |
Long Context Llama 3-8B leads
Llama 3-8B: 34.2 (#246), Llama 4 Maverick: 31.4 (#274)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| LMArena Longer Query | 1128 | 1280 |
| Fiction.LiveBench | — | 46.2% |
Writing & Preference Llama 4 Maverick leads
Llama 3-8B: 37.7 (#251), Llama 4 Maverick: 39.0 (#247)
| Benchmark | Llama 3-8B | Llama 4 Maverick |
|---|---|---|
| LMArena Text | 1166 | 1287 |
| LMArena Creative Writing | 1150 | 1266 |
| LMArena Multi-Turn | 1152 | 1289 |
| Short-Story Creative Writing | — | 62% |
| EQ-Bench Creative Writing | — | 860 |
| WildBench | — | 80% |
Frequently asked questions
Is Llama 3-8B better than Llama 4 Maverick?
Llama 4 Maverick is the stronger model overall, scoring 30.6 to 25.3 on the Noometry Index. Llama 3-8B costs 2.2× less per token, which makes it the better buy when Llama 4 Maverick's lead doesn't matter for your workload.
Which is cheaper, Llama 3-8B or Llama 4 Maverick?
Llama 3-8B is cheaper. It lists at $0.14 per million input tokens and $0.14 per million output tokens; Llama 4 Maverick lists at $0.19 and $0.65.
Is Llama 3-8B or Llama 4 Maverick better for coding?
Llama 3-8B scores higher on coding benchmarks: 30.5 versus 26.3 in the Noometry coding category.
Which has the bigger context window?
Llama 4 Maverick does, with 128K tokens against 8K.
How many benchmarks do Llama 3-8B and Llama 4 Maverick share?
25 benchmarks have published results for both models. Llama 3-8B has 34 scored results on Noometry and Llama 4 Maverick has 54.