兔子API 兔子API
中

provider

Meta Llama

Direct answerMeta's open-weight model family, available for self-hosting or through partner cloud platforms.

Updated · Reviewed

Separate the model originator from inference services

Meta Llama is an open-weight model family, not a uniform public inference REST API operated by Meta. The official repository, per-generation model card, Community License, and Acceptable Use Policy define weights, capabilities, and use. AWS, Azure, local runtimes, and other hosts separately define base URL, authentication, SLA, and data policy.

This site's route is a hosting-channel protocol

The listed /v1/chat/completions route is this gateway's compatibility interface to an actual Llama hosting channel. It is not a Meta direct endpoint and does not imply an api.meta.com. Confirm the full model ID in the model marketplace; response fields, tools, vision, streaming, and billing follow the selected channel adapter.

Pin the exact weight and runtime configuration

“Llama” identifies only a family. Generation, Base versus Instruct, parameter size, vision support, official versus community derivative, quantization, and fine-tuning all change behavior. Record the weight version or digest, model card, prompt template, tokenizer, context, quantization, and inference engine. Do not migrate using only a short alias.

Open weights do not mean unconditional use

Each generation uses its own custom Community License and is subject to an Acceptable Use Policy. Commercial use, redistribution, derivative naming, attribution, or very large-scale use can carry specific conditions; community quantizations and fine-tunes can add other licenses. Base legal and compliance review on the exact weight package and accompanying terms.

Self-hosting transfers the operating burden

Self-hosting teams own GPU and memory capacity, batching, concurrency, timeouts, upgrades, telemetry, safety filtering, and abuse response. For a managed channel, verify region, log retention, content safety, quota, and retirement policy. In either case, evaluate only the languages, modalities, and context documented by the exact model card.

Reviewed provider catalog

Latest models by capability

These are provider-catalog models, not a promise of availability on this site. Confirm callable IDs, endpoints, and pricing in the model marketplace.

Text, reasoning, and code

Check live availability in the model marketplace

Use cases

  • Open-weight deployment
  • Multimodal understanding
  • Broad cloud and tooling ecosystem

API protocols

  • /v1/chat/completions

FAQ

Does Meta Llama have one official public API base URL?

No single Meta-operated public inference REST service covers every Llama weight. Choose a self-hosted runtime or actual cloud host and follow that host's authentication, paths, and response contract.

Is this site's /v1/chat/completions a Meta API?

No. It is a compatibility protocol backed by an actual hosting channel, not an endpoint operated by Meta. Model IDs, context, billing, and data policy depend on that channel.

Do open Llama weights allow unconditional commercial use?

No. Each generation has its own Community License and Acceptable Use Policy, potentially covering attribution, redistribution, branding, scale, and prohibited uses. Read the exact release.

Why can the same Llama name behave differently across platforms?

Hosts can use different weight snapshots, quantization, prompt templates, tokenizers, context settings, and inference parameters. Pin the full identifier and run contract and quality regression tests.

  • Meta LlamaLlama is Meta's open-weight family, normally used through self-hosting or cloud partners; model cards, licenses, prompt templates, and the actual runtime jointly define behavior.
  • Choosing and calling coding modelsSelect coding models by patch correctness, test pass rate, tool success, latency, and cost on real repositories, not marketplace copy or a single benchmark.
  • Tool use and AI agent model guideAn agent is a controlled loop, not merely tool calling: structured schemas, least privilege, state, idempotency, timeouts, approval, audit, and replayable evaluation.
  • AI API quickstart: from API key to first requestAn AI API call needs a matching base URL, API key, model ID, and endpoint; start with minimal curl, then add error handling, cost controls, and key security.
  • Common AI API error codes and troubleshootingHandle AI API failures by parsing the HTTP status, provider code, error type, message, retry hints, and request ID together. Fix request, authentication, billing, and hard-quota causes first; retry only transient throttling, network failures, and selected 5xx errors with strict bounds.

Official sources

  1. Meta Llama Get Started Official
  2. Meta Llama Models Repository Official
  3. Llama 4 Model Card Official
  4. Llama 4 Community License Official