Enterprise DNA Enterprise DNA
Meta logo

Meta

United States

The lab that made open-weight AI mainstream, with the Llama family teams build on and self-host worldwide.

HQ
Menlo Park, California
Founded
2013
Runs on
Self-hostable everywhere, and hosted by every major cloud and inference provider
Structure
A division of Meta Platforms

Capability fingerprint

Best model per axis, vs the whole field

Intelligence 50.6 Coding 71.3 Agentic 37.5 Speed 122 tps Value 132.2 pts/$ Context 10M Openness 87% open

Leads on Context and Coding

House style

Llama is the open-weight default. The models are solid all-rounders that anyone can download, fine-tune, and run on their own hardware, which made Llama the base layer for a huge amount of the open ecosystem. The house style is accessibility and ecosystem over chasing the top of the closed leaderboards.

Naming, decoded

Llama is the family, with version numbers marking generations (2, 3, 3.1, 3.3, 4) and parameter sizes (8B, 70B, 405B) marking scale. Named editions like Scout and Maverick denote specific Llama 4 variants.

The Enterprise DNA take

Llama is the ownership play. When self-hosting, fine-tuning, or full data control matters more than topping the leaderboard, it is the default, and the ecosystem around it is unmatched. For raw frontier scores you look elsewhere, but for control you start here.

How we use it: Enterprise DNA treats Llama as the reference open-weight base for any Omni workload that needs to run on a customer's own infrastructure.

Strengths

  • +Fully open weights, the base of the open ecosystem
  • +Self-host anywhere for complete data control
  • +Massive tooling, fine-tune, and community support

Watch-outs

  • Top-tier scores trail the leading closed models
  • You own the hosting and ops burden when self-hosting
  • License has conditions at very large scale

Best for

Teams that want to own their stack, fine-tune on their own data, and self-host for control or cost, without depending on any single vendor's API.

Pick a model

Release timeline

  1. Muse Spark 1.1 2026-07-16 50.6 intel
  2. Prompt Guard 2 86M 2025-05-29
  3. Llama Guard 4 12B 2025-04-30
  4. Llama 4 Maverick 2025-04-05 14.3 intel
  5. Llama 4 Scout 2025-04-05 10.0 intel

Every Meta model

52 models tracked · 45 open / 7 closed · sortable and searchable, prices refreshed nightly.

Showing 52 of 52 models · ranked by Intelligence Index

Model
Meta logo Muse Spark 1.1 Meta 50.6 71.3 37.5 122 $1.25 $4.25 1.0M 10d ago
Meta logo Llama 4 Maverick Meta 14.3 16.3 1.3 113 $0.200 $0.800 1.0M 1y ago
Meta logo Llama 4 Scout Meta 10.0 8.2 1.1 88 $0.100 $0.300 10M 1y ago
Meta logo Llama 3.3 70B Instruct (free) Meta 9.4 11.9 0.3 · $0 $0 131K 1y ago
Meta logo Llama 3.3 70B Instruct Meta 9.4 11.9 0.3 · $0.130 $0.400 131K 1y ago
Meta logo Llama 3.1 8B Instruct Meta 7.6 5.4 0.5 · $0.050 $0.080 131K 2y ago
Meta logo esm2-650m Meta · · · · $0 $0 128K 1y ago
Meta logo esmfold Meta · · · · $0 $0 128K 2y ago
Meta logo Llama 3.1 70b Instruct Meta · · · · $0 $0 128K 2y ago
Meta logo Llama 3.1 8B Instruct Meta · · · · $0 $0 16K 1y ago
Meta logo Llama-3.2-11B-Vision-Instruct Meta · · · · $0 $0 128K 1y ago
Meta logo Llama 3.2 1b Instruct Meta · · · · $0 $0 128K 1y ago
Meta logo Llama 3.2 3B Instruct Meta · · · · $0 $0 33K 1y ago
Meta logo Llama-3.2-90B-Vision-Instruct Meta · · · · $0 $0 128K 1y ago
Meta logo Llama-3.3-70B-Instruct Meta · · · · $0 $0 128K 1y ago
Meta logo Llama-3.3-70B-Instruct Meta · · · · $0 $0 128K 1y ago
Meta logo Llama 4 Maverick 17B 128E Instruct FP8 Meta · · · · $0 $0 128K 1y ago
Meta logo Llama-4-Maverick-17B-128E-Instruct-FP8 Meta · · · · $0 $0 128K 1y ago
Meta logo Llama 4 Maverick 17b 128e Instruct Meta · · · · $0 $0 128K 1y ago
Meta logo Llama 4 Scout 17B 16E Instruct Meta · · · · $0 $0 128K 1y ago
Meta logo Llama-4-Scout-17B-16E-Instruct-FP8 Meta · · · · $0 $0 128K 1y ago
Meta logo Llama Guard 4 12B Meta · · · · $0 $0 128K 1y ago
Meta logo Llama 3.2 3B Instruct (free) Meta · · · · $0 $0 131K 1y ago
Meta logo Meta-Llama-3.1-405B-Instruct Meta · · · · $0 $0 128K 2y ago
Meta logo Meta-Llama-3.1-70B-Instruct Meta · · · · $0 $0 128K 2y ago
Meta logo Meta-Llama-3.1-8B-Instruct Meta · · · · $0 $0 128K 2y ago
Meta logo Meta-Llama-3-70B-Instruct Meta · · · · $0 $0 8K 2y ago
Meta logo Meta-Llama-3-8B-Instruct Meta · · · · $0 $0 8K 2y ago
Meta logo Llama Guard 3 8B Meta · · · · $0.020 $0.060 131K 2y ago
Meta logo Llama 3.2 1B Instruct Meta · · · · $0.027 $0.201 131K 1y ago
Meta logo Llama Prompt Guard 2 22M Meta · · · · $0.030 $0.030 512 1y ago
Meta logo Llama 3 8B Instruct Meta · · · · $0.040 $0.040 8K 2y ago
Meta logo Prompt Guard 2 86M Meta · · · · $0.040 $0.040 512 1y ago
Meta logo Llama 3.2 3B Instruct Meta · · · · $0.051 $0.335 131K 1y ago
Meta logo Llama 3.2 1B Instruct Meta · · · · $0.100 $0.100 128K 1y ago
Meta logo Llama 4 Scout 17B 16E Meta · · · · $0.110 $0.340 131K 1y ago
Meta logo Meta Llama 3 8B Instruct Lite Meta · · · · $0.140 $0.140 8K 2y ago
Meta logo Llama 3.2 3B Instruct Meta · · · · $0.150 $0.150 128K 1y ago
Meta logo Llama 3.2 11B Vision Instruct Meta · · · · $0.160 $0.160 128K 1y ago
Meta logo Llama Guard 4 12B Meta · · · · $0.180 $0.180 164K 1y ago
Meta logo Llama 3.1 8B Instruct Meta · · · · $0.220 $0.220 128K 2y ago
Meta logo Llama 4 Maverick Instruct Meta · · · · $0.270 $0.850 1.0M 1y ago
Meta logo Llama 3.2 11B Vision Instruct Meta · · · · $0.345 $0.345 131K 1y ago
Meta logo Llama 4 Maverick 17B 128E Instruct Meta · · · · $0.350 $1.15 524K 1y ago
Meta logo Llama 3.2 90B Vision Instruct Meta · · · · $0.350 $0.400 16K 1y ago
Meta logo Llama 3.1 70B Instruct Meta · · · · $0.400 $0.400 131K 2y ago
Meta logo Llama3 70B Instruct Meta · · · · $0.510 $0.740 8K 2y ago
Meta logo Llama 3.1 70B Instruct Meta · · · · $0.720 $0.720 128K 2y ago
Meta logo Llama 3.2 90B Vision Instruct Meta · · · · $0.720 $0.720 128K 1y ago
Meta logo Llama 3.3 70B Instruct Meta · · · · $0.720 $0.720 128K 1y ago
Meta logo Llama 3.3 70B Meta · · · · $1.04 $1.04 131K 1y ago
Meta logo Llama 3.1 405B Instruct Turbo Meta · · · · $3.50 $3.50 128K 2y ago

Intelligence, Coding, Agentic scores and Speed (median output tokens/sec) from Artificial Analysis. Pricing, context and release dates from the OpenRouter catalogue, refreshed nightly. A dash means the metric isn't published for that model.

Meta did something the other frontier labs did not: it gave the weights away. The Llama family, built by Meta’s FAIR and GenAI teams, made open-weight AI mainstream and became the base layer that a huge share of the open ecosystem fine-tunes, hosts, and builds on.

The house style is accessibility over leaderboard chasing. Llama models are solid all-rounders that anyone can download, adapt, and run on their own hardware, which is exactly the point. When self-hosting, fine-tuning, or full data control matters more than the last few points of benchmark, Llama is the default and the ecosystem around it is unmatched.

The catalogue below is every Llama model we track, joined live to pricing, context, and benchmark scores.