Llama 4 Maverick
Meta's high-performance open-weight MoE model — 400B parameters of knowledge at ~17B of compute per token.
About
Llama 4 Maverick is the top tier of Meta's Llama 4 open-weight family: a 400B-total-parameter Mixture-of-Experts model with only 17B active parameters per token and native text+image multimodality.
As of Oct 2026 it remains Meta's newest Llama release (April 2025); Meta's 2026 frontier work has shifted to the proprietary Muse line and the newer open-weight Muse Glimmer (30B, Apache 2.0, Aug 2026). There is no first-party Meta inference API — developers self-host or use third-party providers. Benchmark figures below are April-2025 launch numbers and should be treated as directional today.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Meta-reported (April 2025 launch era — directional today):
- MMLU: 85.5
- MATH: 84.8
- HumanEval: 82.3
- MMMU: 80.5
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Meta is established and the open weights plus massive community ecosystem (vLLM, Ollama, llama.cpp) make this the most deployable open model in the catalog. Caveats: the benchmark story is 18 months old, there is no first-party API or pricing, the Llama Community License is more restrictive than Apache 2.0, and Meta's frontier attention has moved to the proprietary Muse line.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.1/10
- Maintenance · recency, releases, issue load
- 5.0/10
- Openness · license + docs
- 7.2/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · HF top match: meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8 (community upload counts as proxy where official repo is gated)
- · HF downloads 43,853 (peer-normalized)
- · released 556 days ago
- · manual review: benchmarks from Apr 2025; restrictive license