Rumored and announced LLM releases

OpenAI's mid-2026 flagship preview, headlined by an industry-leading 1.5M-token context window and long-horizon agentic tool use.

Access

FrontierProprietary

Context

1.5M

Parameters

Undisclosed

Source

open

Google DeepMind model announced at Google I/O 2026 but not released at the event. Sundar Pichai told attendees to wait until the following month; Google positioned it as the most powerful Gemini model in the works.

Access

FrontierProprietary

Context

2M

Parameters

Undisclosed

Source

open

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.

Access

FrontierOpen weights

Context

—

Parameters

2T · 288B active

Source

open

The public, guardrailed sibling of Mythos and Anthropic's first Mythos-class model for general use. It launched in June 2026 and was then pulled under a U.S. government intervention that restricted Fable and Mythos access.

Access

FrontierProprietary

Context

—

Parameters

Undisclosed

Source

open

OpenAI's limited-preview cybersecurity model for vetted defenders in Trusted Access for Cyber. It is tuned for more permissive authorized security workflows such as vulnerability triage, patch validation, malware analysis, red teaming, and controlled exploit validation, but is not generally available.

Access

FrontierProprietary

Context

—

Parameters

Undisclosed

Source

open

A frontier model Anthropic disclosed but declined to release publicly, citing security risk. Access was limited through Project Glasswing and later suspended during the Fable/Mythos government intervention.

Access

FrontierProprietary

Context

—

Parameters

Undisclosed

Source

open

A speculative Z.ai frontier model tracked after Z.ai founder Jie Tang responded to Elon Musk's prediction of a Chinese Fable 5-class model by saying it would not take that long. Name, architecture, weights, and launch timing remain unconfirmed.

Access

FrontierProprietary

Context

—

Parameters

Undisclosed

Source

open

Meta's reported next frontier model after Llama 4, tracked under the internal codename Avocado. Reports say the model was delayed after internal trials lagged the strongest Google, OpenAI, and Anthropic models; access, license, and final branding remain unconfirmed.

Access

FrontierProprietary

Context

—

Parameters

Undisclosed

Source

open

Rumored successor to DeepSeek R1. Reports say development and launch timing were affected by hardware constraints around Huawei Ascend training and Nvidia availability; final specs, license, and release date remain unconfirmed.

Access

FrontierOpen source

Context

—

Parameters

Undisclosed

Source

open

Rumored next major Grok model. Elon Musk said after GPT-5's launch that Grok 5 would arrive before the end of 2025, but no broad public Grok 5 release is logged in this catalog yet.

Access

FrontierProprietary

Context

—

Parameters

Undisclosed

Source

open