LLM Releases

Source packet

Collaboration kit

Latest LLM release source packet preview

A compact research packet for journalists, newsletter writers, creators, and analysts who need recent model-release facts with links back to the original sources.

Request source materialsRequest API or data accessView release radar
12
Events
9
Releases
0
Updates
8
Labs

Copyable brief

What changed recently

In the latest 30-day LLM Releases window, the catalog tracks 12 model lifecycle events across 8 labs, including 9 releases and 0 updates. 8 preview entries include original source links. The window is anchored to Jun 18, 2026, so the briefing stays stable between crawls.

This public page shows a sample of the packet structure. For a full source packet, API access, a custom model slice, or embargo-friendly collaboration, send a request through the partner form.

Suggested citation

LLM Releases, "Latest LLM release source packet preview," https://llm-releases.com/source-packets/latest. Accessed 2026-06-20.

Original links

Sample release events

Previewing 8 of 12 tracked events.

  1. ReleasedMoonshot releases Kimi K2.7 Code
    Moonshot AI

    Open coding-focused Kimi model with 1T total / 32B active parameters, native image/video input, and always-on thinking mode.

    source
  2. ReleasedZ.ai releases GLM-5.2 with a 1M-token context
    Z.ai (Zhipu AI)

    MIT-licensed GLM flagship focused on long-horizon coding, agentic engineering, and IndexShare sparse-attention reuse.

    source
  3. ReleasedMiniMax releases MiniMax-M3
    MiniMax

    Native multimodal 428B/23B-active model with one-million-token context and MiniMax Sparse Attention.

    source
  4. WithdrawnUS government orders Anthropic to pull Fable 5 and Mythos 5
    Anthropic

    Access suspended three days after launch under an export-control directive citing national security.

    source
  5. ReleasedClaude Fable 5 released — first public Mythos-class model
    Anthropic

    Available across the Claude API, AWS, and Microsoft Foundry.

    source
  6. AnnouncedGPT-5.6 announced with a 1.5M-token context window
    OpenAI

    OpenAI previewed GPT-5.6, claiming the largest context window of any frontier model.

    source
  7. ReleasedNVIDIA releases Nemotron 3 Ultra 550B-A55B
    NVIDIA

    Largest Nemotron 3 model appears on NVIDIA NIM with downloadable weights, 1M context, and agentic reasoning positioning.

    source
  8. ReleasedAlibaba releases Qwen3.7-Plus multimodal agent model
    Alibaba (Qwen)

    Lower-cost multimodal sibling of Qwen3.7-Max with text, image, and video input and a 1M-token context.

    source

Related