Llama 4 Maverick by Meta — 400B MoE (17B active) open-weight local LLM. License: Llama 4 Community. Benchmarks, VRAM requirements per quantization, GPU compatibility, pricing and community reviews on slopsome.com.
Key specs
| Type | Local open-weight |
|---|
| Parameters | 400B total · MoE, 17B active |
|---|
| Architecture | MoE (128 experts) |
|---|
| Context window | — |
|---|
| Knowledge cutoff | 2024-08-01 |
|---|
| Modalities | text, image |
|---|
| Recommended backends | — |
|---|
| Minimum viable rig | — |
|---|
Benchmark scores
| GPQA Diamond | 69% |
|---|
| SWE-bench Verified | 43% |
|---|
| AIME | 60% |
|---|
| MMLU-Pro | 81% |
|---|
| BFCL v3 (tool use) | 68% |
|---|
| Composite score | 6.06 |
|---|
| Community rating | No reviews yet |
|---|
VRAM & disk per quantization
| Quant | VRAM | Disk | RAM | Context |
|---|
| Q4_K_M | 205 GB | 200 GB | 256 GB | 1000K |
| Q8 | 405 GB | 400 GB | 448 GB | 1000K |
API pricing (per 1M tokens)
| Provider | Input | Output | Free tier |
|---|
| Groq | $0.2 | $0.6 | No |
| Fireworks AI | $0.22 | $0.88 | No |
| Together AI | $0.27 | $0.85 | No |