New Claude Haiku 3.5 is good at coding, but this price increase vs Haiku 3 is sad. 😞

New Claude 3.5 Haiku comes in 4th on the aider code editing leaderboard with 75%. This is just behind the old 3.5 Sonnet 06/20.

84% Sonnet 10/22
80% o1-preview
77% Sonnet 06/20
75% Claude 3.5 Haiku
47% Claude 3 Haiku

https://t.co/mBVaUPGHPl https://t.co/1jAuSEpMW3

Paul Gauthier

summary_large_image

Claude 3.5 Haiku is now available on our API, Amazon Bedrock, and Google Cloud's Vertex AI.

Haiku is fast and particularly strong at coding. It outperforms state-of-the-art models—including GPT-4o—on SWE-bench Verified, which measures how models solve real software issues. https://t.co/SR0i9W7W0T

During final testing, Haiku surpassed Claude 3 Opus, our previous flagship model, on many benchmarks—at a fraction of the cost.

As a result, we've increased pricing for Claude 3.5 Haiku to reflect its increase in intelligence: https://t.co/UWTd1RDRVE.

Anthropic

New Claude Haiku 3.5 is good at coding, but this price increase vs Haiku 3 is sad. 😞 

https://x.com/paulgauthier/status/1853532441077850319?s=46&t=jQHN-g4HjwTKZe7EhUgxHg

https://x.com/anthropicai/status/1853498270724542658?s=46&t=jQHN-g4HjwTKZe7EhUgxHg

I’m a builder looking for the next challenge. 

My guess is that they are afraid of cannibalization vs new sonnet and drop in revenue (they had some hiccups with funding)  and they have a problem with capacity (they’ve downgraded free Claude access to Haiku 3 last week).

But it’s a very bad strategic move. They could have increase their market share. 
With new Open AI releases this year this will hurt them a lot.

Bro I don’t even think it’s actually much better than Opus

follow @1dolinski --

challenge your habits and share --

https://t.me/aiexecutive_1dolinski_bot

Price is the only reason , that’s why I’m disappointed. 

It would make a difference for AI products, especially agents and others that require bigger contexts. 

I’ve been tinkering with one for ages that is still waiting for a super cheap model but with very good reasoning and coding capabilities (and big context window).