Meta AI model releases

Meta Superintelligence Labs' first paid model, released July 9, 2026 in US public preview on the Meta Model API. A natively multimodal reasoning model (text, image, video, PDF, and audio input; text output) with explicit chain-of-thought reasoning and a 1M-token context window that the model actively compacts. Positioned for agentic work and coding — tool use, multi-step workflow coordination, and long-horizon autonomous tasks — and pitched by Meta as roughly a quarter of the price of comparable Anthropic and OpenAI models at $1.25 in / $4.25 out per Mtok (with $20 in free credits per new API account). Marks the first time Meta has charged businesses for one of its models, a departure from the open-weight Llama strategy. Closed weights, undisclosed size. Vendor and third-party benchmarks place it around the Opus 4.8 / GPT-5.5 tier — strongest as an agent/workflow model and in tool-augmented reasoning, competitive but not dominant on coding and multimodal tasks.

Meta's new frontier model behind Meta AI for U.S. users, identified in reporting as the public release of the former Avocado effort and positioned to compete with Gemini, GPT, and Claude on multimodal assistant tasks.

Historical rumor/codename row retained for provenance. Recent reporting identifies the public productized model as Muse Spark, now tracked separately in the catalog.

Meta's flagship open-weight MoE; highest MMLU among open models at release.

Efficient open-weight MoE designed for very long context on modest hardware.

Meta's announced but unreleased Llama 4 teacher model: a multimodal MoE with 288B active parameters and nearly 2T total parameters. Meta says it was still training when Scout and Maverick shipped and that those released models were distilled from Behemoth.

Late-2024 70B Llama update delivering much of the 405B instruction-following quality at lower serving cost.

First Llama family release with native vision models, alongside smaller edge-oriented 1B and 3B text models.

Meta's first frontier-scale open Llama model, with 405B parameters, 128K context, multilingual support, and tool-use improvements.

First Llama 3 release, with 8B and 70B open models and a stronger tokenizer, data mix, and post-training stack.

Meta's first code-specialized Llama model family, released in base, Python, and instruction-tuned variants.

The release that made capable open-weight models genuinely usable for production.

Meta's first LLaMA, released to researchers; its leak catalyzed the open-weight movement.

A science-focused model whose public demo was withdrawn after just three days over confidently wrong outputs — an early, instructive retraction.