Llama 3.1 405B
The 405B giant: Meta's biggest open-weights model, GPT-4-class.
About
Llama 3.1 405B Instruct was Meta's landmark July 2024 release: the first open-weights model at GPT-4-class capability — 405B dense parameters, 128K-token context, 8 languages, tool use. Weights are gated (manual approval) under the Llama 3.1 Community License — custom, non-OSI (acceptable-use policy, 700M-MAU clause): "open weights", never "open source". The practical reality: ~810GB of weights means almost nobody self-hosts it — hosted endpoints (or the 70B sibling) are the sane path. Historically pivotal as the model that proved open weights could rival frontier labs; superseded by Llama 3.3 and 4.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported (Meta, Jul 2024): GPT-4-competitive on MMLU, HumanEval, MATH, GPQA — provider-reported. Historical; the frontier moved on.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (verified tier, metrics-v1). Evidence: HF downloads 10,098 (peer-normalized); released 812 days ago. Manual review: biggest open-weights dense model of 2024 (405B, GPT-4-class per Meta); ~810GB weights = almost nobody self-hosts; superseded by 3.3/4. License: Llama 3.1 Community License — custom, non-OSI (700M MAU clause, acceptable-use policy); describe as 'open weights', never 'open source'.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.2/10
- Maintenance · recency, releases, issue load
- 4.0/10
- Openness · license + docs
- 7.2/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · HF downloads 10,098 (peer-normalized)
- · released 812 days ago
- · manual review: biggest open-weights dense model of 2024 (405B, GPT-4-class per Meta); ~810GB weights = almost nobody self-hosts; superseded by 3.3/4