Ctrl K
Back to models

Meta

Llama 4 Maverick

  • Released
  • Gated weights
  • Llama 4 Community License Agreement
Compare
Intelligence Index

Standard setting

9.3#73 of 83measured models
Measured · Artificial Analysis v4.3.2
API price from
$0.20 in / $0.80 out / 1M tokOpenRouter · DeepInfra · FP8
Context window
1M tok

Performance profile

ReasoningRank #80 of 107
67.1%
GPQA Diamond
CodingRank #79 of 94
31.7%
SciCode
Cost efficiencyRank #16 of 67
242.8pts / $
Intelligence ÷ eval cost

Capabilities

Context window

Modalities

Measured setting

Integration

Reasoning trade-offs

Source (opens in a new tab)
Intelligence Index v4.3.2
Llama 4 Maverick
Cost per task (USD) · log scale
Llama 4 MaverickStandard
Intelligence Index
9.3
Cost / task
$0.0383
Evaluation time
36.0s

Pricing & access

API route
OpenRouter · DeepInfra · FP8API
Tier & contextStandard

USD per 1M tokens

Input
$0.20
Output
$0.80

Run locally

GGUF
Context
Estimated memory

~250.2 GiB

RAM + VRAM

Estimate assumptions

One text chat at the selected context, with a 16-bit cache and app allowance. Operating system and image processing are extra; optimized caches can use less.

256 GiB RAMwithout GPU offload

Memory budgetOne text chat
Weights
226.1 GiB
Context cache
1.5 GiB16-bit cache
App allowance
22.6 GiB
Download
242.8 GB5 files · keep together+1.7 GB for images (opens in a new tab)
Disk space
~274 GBIncludes image file
Available software
Ollama (opens in a new tab)LM Studio (opens in a new tab)Jan (opens in a new tab)KoboldCpp (opens in a new tab)llama.cpp (opens in a new tab)
Get GGUF (opens in a new tab)
Set up in Ollama · Q4_K_M · 8K context

Save the following as Modelfile. The context matches the estimate above.

FROM hf.co/unsloth/Llama-4-Maverick-17B-128E-Instruct-GGUF:Q4_K_M
PARAMETER num_ctx 8192

Then run these commands in your terminal:

ollama create llmcurrent-local -f Modelfile
ollama run llmcurrent-local

Ollama import instructions (opens in a new tab) · For LM Studio or Jan, open the app’s model search, enter unsloth/Llama-4-Maverick-17B-128E-Instruct-GGUF, choose Q4_K_M, and set context to 8,192 tokens in the app’s model settings. Software links above open each app’s setup guide.

Images need extra memoryvLLM setup (opens in a new tab)

Benchmarks

Full leaderboardSource (opens in a new tab)

Humanity’s Last Exam

Frontier reasoning

Llama 4 Maverick · Standard#111 of 126

  1. 1Claude Opus 5.5Max61.4%
  2. 2Claude Fable 5.1Max59.1%
  3. 3Gemini 4 ArgonHigh57.1%
  5. 108Tiny Aya GlobalStandard5.2%
  6. 109Granite 4.0 MicroStandard5.0%
  7. 110Phi-4 Multimodal InstructStandard5.0%
  8. 111Llama 4 MaverickStandard4.9%
  9. 112NVIDIA Nemotron Nano 9B V2Reasoning4.9%
  10. 113Nemotron 3 Nano Omni 30B A3B ReasoningReasoning4.8%
  11. 114Qwen3 Omni 30B A3B InstructStandard4.6%

Peers use their best measured result on this benchmark.

Alternatives

Each model at its highest-scoring measured setting · Intelligence Index v4.3.2 and its evaluation cost per task.