All models
Meta
Llama 4 Maverick fastruns locally
meta/llama-4-maverick
Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...
Input / 1M
$0.35
USD per million tokens
Output / 1M
$1.00
USD per million tokens
Est. cost / token
$0.000000512
$0.51 per 1M, 3:1 mix
Context window
524K
max output 33K
Major features
VisionOpen weightsCachingJSON schemaTools
Task suitability (heuristic 1–5, not a benchmark)
- coding3
- reasoning2
- math2
- summarization4
- extraction4
- classification4
- chat3
- creative3
- translation4
- vision3
- long context4
- agentic3
Run it locally
Will it run on my machine? →ollama run llama4:maverick| Quantization | Quality | Memory needed (8K ctx) |
|---|---|---|
| Q2_K | poor | 182 GB |
| Q3_K_M | fair | 211 GB |
| Q4_K_Mrecommended | good | 263 GB |
| Q5_K_M | very good | 305 GB |
| Q6_K | excellent | 351 GB |
| Q8_0 | near-lossless | 452 GB |
| F16 | lossless | 846 GB |
Estimates. Mixture-of-experts: memory covers all parameters; only the active ones are used per token, so it runs faster than its size suggests.
Price history
| Recorded | Input | Output | Cached | Source |
|---|---|---|---|---|
| 2026-11-05 03:20 | $0.35 | $1.00 | $0.17 | openrouter |
A row is added whenever a sync detects a price change. Vendor pricing source
Pricing (USD per 1M tokens)
- Input
- $0.35
- Output
- $1.00
- Cached input
- $0.17
- Blended (3:1)
- $0.51
Details
- Company
- Meta
- Version
- 4
- Family
- Llama
- Released
- 2025-04-05
- Parameters
- 400B (17B active)
- Max output
- 33K
- Open weights
- Yes
- Inputs
- image, text
- Last synced
- 46 min ago