#Why the answer is Claude 3.5 Sonnets

10 messages · Page 1 of 1 (latest)

prisma bear
#

ask What version of Claude am I talking to?

image.png
covert lilyBOT
#

Hi @prisma bear. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

The issue where Warp responds with "Claude 3.5 Sonnet" when you have selected "Claude 4 Sonnet" as your model appears to be a bug that has been reported by users. Specifically, users have observed that even after choosing Claude 4 Sonnet, the agent identifies itself as Claude 3.5 Sonnet in its responses. This has been documented in a recent GitHub issue, where the agent replied, "I am Claude 3.5 Sonnet, not Claude 4" despite the user selecting Claude 4 Sonnet in the model picker. This suggests that either the model selection is not being respected internally, or the agent is misreporting its identity in responses. The issue is acknowledged as a regression and is currently open for investigation by the Warp team Model misidentification issue.

If you are experiencing this, it is not due to your configuration but rather a known issue in the current version of Warp. You may want to follow the linked GitHub issue for updates or try updating Warp if a fix is released.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@prisma bear Kapa.ai is still learning and improving, please let me know how I did by reacting below.

prisma bear
#

<@&1136444479654600734>

brisk cosmos
#

models dont know who they are unless it is explicitly stated in the system prompt they are prone to hallucinate

coral parrot
#

@prisma bear Gus is correct, and we dont indicate the version in the system prompt. This is why claude may report another version. Just trust that the model you pick is the one we provide.

brisk cosmos
#

i mean think about it logically, it'd be quite easy to just serve you a cheap model and then include a name in the system prompt anyway so its not a good argument either way