Rotary Position Embedding (RoPE)

Deployment timeline

1. Llama 4 Scout

   Deployed 2025-04-05 · Velocity 4y

   “Based on Llama 3 architecture which uses RoPE; Llama 4 Scout is a direct evolution.”

   high
2. Llama 4 Maverick

   Deployed 2025-04-05 · Velocity 4y

   “Llama family models consistently use RoPE. Llama 4 is a direct successor.”

   high
3. Claude Opus 4.6

   Deployed 2026-02-16 · Velocity 5y

   “Anthropic's research mentions using rotary position embeddings (RoPE) in transformer architectures.”

   medium
4. GPT-4o

   Deployed 2026-02-16 · Velocity 5y

   “GPT models since GPT-3 use rotary position embeddings (RoPE). GPT-4o's architecture is a direct evolution.”

   high
5. GPT-5

   Deployed 2026-02-16 · Velocity 5y

   “RoPE is a standard positional encoding used in modern Transformer LLMs, including GPT series.”

   high
6. Claude 3

   Deployed 2026-02-18 · Velocity 5y

   “Claude 3 uses Rotary Position Embeddings (RoPE) for positional encoding, per technical details.”

   high
7. Gemini 3 Pro

   Deployed 2026-02-19 · Velocity 5y

   “Gemini models use rotary position embeddings (RoPE).”

   high
8. Gemini 3.1

   Deployed 2026-02-20 · Velocity 5y

   “Gemini models use Rotary Position Embeddings (RoPE) for position encoding.”

   medium
9. Claude 3.5 Sonnet

   Deployed 2026-02-23 · Velocity 5y

   “Anthropic's Claude models use rotary position embeddings (RoPE) for position encoding.”

   high
10. Claude Haiku 4.5

    Deployed 2026-02-25 · Velocity 5y

    “Claude models use rotary position embeddings (RoPE) for positional encoding.”

    high
11. GPT-5.3

    Deployed 2026-02-26 · Velocity 5y

    “RoPE is a standard position encoding in modern LLMs; GPT-5.3 likely uses it for better length extrapolation.”

    medium
12. Claude 4.5

    Deployed 2026-02-26 · Velocity 5y

    “Anthropic's Claude models use rotary position embeddings (RoPE) for positional encoding.”

    medium
13. Gemini 3 Flash

    Deployed 2026-02-27 · Velocity 5y

    “Gemini models use rotary position embeddings (RoPE), as confirmed in the Gemini 1.5 technical report.”

    high
14. GPT-OSS-120B

    Deployed 2026-03-02 · Velocity 5y

    “As a large language model in the GPT lineage, it almost certainly uses Rotary Position Embedding (RoPE), which is standard in modern transformer architectures.”

    medium
15. Grok 4.20

    Deployed 2026-03-02 · Velocity 5y

    “Grok models are based on the Transformer architecture, which commonly uses RoPE for position encoding.”

    medium
16. Kimi K2.5

    Deployed 2026-03-04 · Velocity 5y

    “Most modern LLMs use RoPE for position encoding; Kimi K2.5's long-context capability aligns with this.”

    medium
17. Gemini 3.1 Flash-Lite

    Deployed 2026-03-05 · Velocity 5y

    “Gemini models use rotary position embeddings (RoPE) for positional encoding.”

    high
18. DeepSeek-V3

    Deployed 2026-03-11 · Velocity 5y

    “DeepSeek-V3 uses Rotary Position Embedding (RoPE).”

    high
19. Mistral Small 4

    Deployed 2026-03-16 · Velocity 5y

    “Mistral models use Rotary Position Embeddings (RoPE).”

    high
20. GLM-5.1

    Deployed 2026-03-21 · Velocity 5y

    “GLM-5.1 uses Rotary Position Embedding (RoPE) for positional encoding.”

    high
21. Qwen 3.6

    Deployed 2026-03-31 · Velocity 5y

    “Qwen models use Rotary Position Embedding (RoPE) for positional encoding.”

    high
22. GPT-5.4-Cyber

    Deployed 2026-04-16 · Velocity 5y

    “GPT models use Rotary Position Embeddings (RoPE) for positional encoding.”

    medium
23. Claude Opus 4.7

    Deployed 2026-04-16 · Velocity 5y

    “Anthropic's Claude 3 model card mentions using rotary position embeddings (RoPE). This is a standard architectural component for their models.”

    high