Meta AIopen-weight
Released: 2025-04-05Llama 4 Maverick
Meta open-weight 400B MoE architecture with 128 experts, activating 17B per token with 1M native context.
Intelligence Index
90.2/ 100
Calibrated multi-domain compositeThroughput Speed
82tok/sec
Output streaming throughputTime To First Token
420ms
Initial chunk server latencyLiveBench Contam-Free
77.1%
September 2026 suite scoreLiveBench Multi-Domain Evaluation Breakdown
Monthly refreshed contamination-free test setsLogical Reasoning88.2%
Coding & Repo Repair79.0%
Mathematics (AIME)89.5%
Data Analysis & Tables82.0%
Language & Comprehension85.5%
Instruction Following90.0%
Sourced Benchmark Evaluations (3)
Verified performance across authoritative benchmarks with provenance tracking.Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run llama-4-maverick
# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "llama-4-maverick",
"messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
}'Evaluation & Sourcing Notes
Scores listed for Llama 4 Maverick represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.