> ## Content Index
> Fetch the complete content index at: https://www.stirup.si/llms.txt
> Use this file to discover other available public pages before exploring further.

# 2026 Model Releases and Access Restrictions by Major Providers
- URL: https://www.stirup.si/2026-model-releases-and-access-restrictions-by-major-providers-bc5a8f22/
- Published: 2026-11-15T12:57:51.000Z
- Updated: 2026-11-15T12:57:51.000Z
- Description: In 2026, multiple providers including Anthropic, DeepSeek, Google, and OpenAI released new text models, with some models restricted to approved organisations as noted by nhimg.org.
- Author: Satish Das
- Tags: News, Stirup Research

## Model Releases in 2026 by Provider

Between July and mid-November 2026, multiple providers released new text models, as confirmed by their official release notes, changelogs, or announcement pages. According to [nhimg.org](https://nhimg.org/llm-release-tracker?ref=stirup.si), Anthropic, DeepSeek, Google, Meta, Mistral AI, OpenAI, and xAI all launched new models during this period. Access to certain models, such as Anthropic’s Mythos series and OpenAI’s cyber and Rosalind models, is restricted to approved, verified, or programme participant organisations, as noted in [nhimg.org](https://nhimg.org/llm-release-tracker?ref=stirup.si). The tracker attributes access status consistently across all providers, ensuring transparency. Each entry includes the API ID and access status at the time of release, as described in [nhimg.org](https://nhimg.org/llm-release-tracker?ref=stirup.si). These details are drawn directly from the sources and are not inferred or modified.

## Deprecation and Retirement of Models

As new large language models (LLMs) continue to emerge in 2026, the deprecation and retirement of older models has become a critical consideration for developers. For example, DeepSeek-V4-Flash transitioned from public beta on 2026-07-31 to retirement on 2026-09-10, while Gemini 3.7 Flash was deprecated eight weeks after its release. According to the LLM Deprecation and Retirement Tracker, developers are advised to check this resource before building on any new model ID to avoid disruptions. Preview models, as noted by OpenAI, may be retired with as little as two weeks' notice. Additionally, image, audio, and embedding models are typically only listed in the tracker when they share a log entry with a text model family or are part of a major family launch.

## Access Status and Restrictions

Access to certain large language models (LLMs) in 2026 remains restricted to approved, verified, or programme participant organisations, as stated by their respective providers. For example, Anthropic’s Mythos models and OpenAI’s cyber and Rosalind models are available only under these conditions. The LLM Release Tracker consistently reflects this access status for all providers, ensuring transparency. Similarly, Moonshot AI’s Kimi K3 series is set to replace the K2 series by the end of December 2026, making K3 dates and IDs critical for users currently on K2\. These restrictions and transitions highlight the importance of tracking access status and model updates for organisations relying on these AI systems.

## Model Release Dates and Tracking

As of 2026, multiple large language models (LLMs) have been released or updated by major providers. OpenAI has announced a knowledge cutoff of 30 April 2026 for GPT-6 Astra and GPT-6.1 Sol, and 1 June 2026 for GPT-6.1 Terra. xAI lists a knowledge cutoff of May 2026 for Grok 4.7\. These dates are based on official announcements from the respective providers. Google Cloud's model versions page provides release dates that may differ from those listed in the Gemini API changelog, with such discrepancies noted as 'Vertex' in the tracking table. Dates are taken directly from the provider's release notes or announcements, ensuring accuracy in tracking model updates. This information helps users align their applications with the most current model capabilities and limitations.

## Model Registration and Best Practices

Model registration and best practices are essential for tracking and managing large language models (LLMs) effectively. According to [nhimg.org](https://nhimg.org/llm-release-tracker?ref=stirup.si), the API id column should be used in configuration, and the release date should be recorded beside it in your model register. [nhimg.org](https://nhimg.org/llm-release-tracker?ref=stirup.si) recommends maintaining a model register that includes id, provider, surface (direct API or cloud), release date, and access status. These practices ensure clarity and consistency in model management. Standards and references are available from various providers, including Anthropic’s Claude Platform release notes, DeepSeek’s API docs change log, Google’s Gemini API release notes, and others. These resources provide detailed information on model updates, lifecycle management, and retirement processes. For further guidance, related NHI Mgmt Group resources such as the AI Model Retirement Guide and Foundation Model Selection Guide are available.

## Sources and reporting notes

- [nhimg.org](https://nhimg.org/llm-release-tracker?ref=stirup.si)

Based on published source material. No independent hands-on testing is claimed.