LLM WORLD

Grok

xAI · 7 tracked releases · November 3, 2023 – September 21, 2026

At a glance

Releases tracked7
First trackedNovember 3, 2023 — Grok-1
LatestSeptember 21, 2026 — Grok 4.7
Span1,053 days
Median gap129 days between releases
Weights1 of 7 open
Record quality5 of 7 verified · how this is judged

Release cadence

One tile per tracked release, placed in the month it shipped, so the pauses in this line are the pauses in the data. Years with nothing tracked keep their row.

  1. 2023
    1. Grok-1, xAI, November 3, 2023
    1
  2. 2024
      0
    1. 2025
      1. Grok 3, xAI, February 19, 2025
      1
    2. 2026
      1. Grok 4.20, xAI, March, 2026Grok 4.20 Multi-agent, xAI, March, 2026
      2. Grok 4.5, xAI, July 8, 2026
      3. Grok 4.6, xAI, August 12, 2026
      4. Grok 4.7, xAI, September 21, 2026
      5

    Lineage

    Ordered by announcement date, and spaced by the gap between releases, so a quiet year looks like one. Lineage is not inferred from version numbers — labs do not number consistently, so the dates decide the order.

    1. Grok-1Text in and out

      coding

    2. +474 days

      Grok 3Text in and outImage inputVideo input

      reasoningvision

      from Grok-1 Context window 8K → 1MWeights Open → Proprietaryfirst evidenced: reasoning, visionfirst evidenced: image in, video in

    3. +375 days

      2 released the same day

      Grok 4.20
      Grok 4.20 Multi-agent

      agentic

    4. +129 days

      Grok 4.5Text in and outImage input

      function callingstructured output

    5. +35 days

      Grok 4.6Text in and outImage input

    6. +40 days

      Grok 4.7Text in and outImage input

    Context window over time

    Only releases with a disclosed context window appear — 2 of 7 are not shown.

    Grok-1 · 2023 8K tokens
    Grok 3 · 2025 1M tokens
    Grok 4.5 · 2026 500K tokens
    Grok 4.6 · 2026 500K tokens
    Grok 4.7 · 2026 500K tokens

    What changed

    Generation to generation, from evidenced values only. A field is compared only where both releases record it — otherwise it is called out as not comparable, so a research gap never reads as a change the lab made.

    Across every generation of this family, parameters could not be compared — the values are not recorded on both sides of any step. That is a gap in this dataset, not a statement about the models. See data quality.

    Grok-1 → Grok 3 474 days

    Context window
    8K → 1M
    Weights
    Open → Proprietary
    Capabilities first evidenced
    + reasoning + vision
    Modalities first evidenced
    + image in + video in

    Grok 3 → Grok 4.20 375 days

    No change in any field both releases record.

    Not comparable here: context window (not disclosed for both releases); capabilities and modalities (not researched on both releases).

    Grok 4.20 → Grok 4.5 129 days

    No change in any field both releases record.

    Not comparable here: context window (not disclosed for both releases); capabilities and modalities (not researched on both releases).

    Grok 4.5 → Grok 4.6 35 days

    No change in any field both releases record.

    Grok 4.6 → Grok 4.7 40 days

    No change in any field both releases record.

    Compare the two most recent side by side →
    All releases from xAI →