3h ago

DeepSeek Agents Identify as Claude Sonnet 4.5 Without Explicit Hints

0
Original post

I tested it with Kimi K2.5 and Qwen 3.6 Plus (default OR models in this TUI) and got the same. Qwen even hallucinated specifically Sonnet 4.5. Inspecting the codebase also suggests the model doesn't get any hints except being a tool-using browser agent. sigh. W for Dario.

12:41 PM · May 23, 2026 View on X

In fairness Qwen *sometimes* identifies as Qwen (of some kind), and Kimi can say it's Codex/GPT-4. Yeah this is specific to DeepSeek, its own identity is nonexistent and it defaults to Claude specifically.

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex

I tested it with Kimi K2.5 and Qwen 3.6 Plus (default OR models in this TUI) and got the same. Qwen even hallucinated specifically Sonnet 4.5. Inspecting the codebase also suggests the model doesn't get any hints except being a tool-using browser agent. sigh. W for Dario.

7:41 PM · May 23, 2026 · 4.7K Views
7:53 PM · May 23, 2026 · 955 Views
DeepSeek Agents Identify as Claude Sonnet 4.5 Without Explicit Hints · Digg