OpenAI's ChatGPT topped Anthropic's Claude Haiku 5.5 by a 3-2 margin in a test of five everyday tasks, while Claude took the two rounds that hinged on tailored advice.
Key Points:
- ChatGPT won the Thanksgiving planning, creative writing and budget calculator rounds.
- Claude Haiku 5.5 won the running shoe and family scheduling rounds with more personal answers.
- Haiku 5.5 replaces Haiku 4.5 in the model menu for Free, Pro and Max users on Oct. 29.
ChatGPT Test Results
Amanda Caswell, a technology journalist who covers AI, tested both chatbots with identical prompts in a comparison published Oct. 15.
Tasks ranged from holiday dinner plans to a household budget calculator. She judged each answer on how closely it followed instructions and how much work it left for her.
ChatGPT won the Thanksgiving round by making nearly every side dish vegan for a 10-person dinner cooked in one oven on a $175 budget. It also opened a dragon story mid-chase with a stronger middle-grade tone, then built a budget tool that users could edit, download and run offline.
Claude won the running shoe round, steering a high-arched runner who logs 35 miles a week toward neutral, max-cushion models. ChatGPT picked a firmer shoe she considered a poorer fit. Claude also won the scheduling test with a more personal plan for a working parent of three facing soccer practice, a post office run and a work deadline.
Also Read: How Apple's iPhone Duo Went From Preorder To Backorder In 30 Minutes
Claude Haiku 5.5 Verdict
Caswell called Haiku 5.5 a good model for everyday questions, but said ChatGPT would probably be her first choice for this mix of tasks. She credited Claude's two wins to advice that felt better suited to her situation.
Anthropic released Haiku 5.5 on Oct. 7 as the smallest and cheapest tier of its Claude lineup. The company says it costs around 75% less to run than Haiku 4.5 and suits quick, repetitive jobs like summaries and live customer support. Customers pay $0.10 per million input tokens, in line with GPT-6 Luna, OpenAI's smallest model.
GPT-6 And Haiku Rollouts
Anthropic is removing Haiku 4.5 from the model menu for Free, Pro and Max users on Oct. 29, leaving Haiku 5.5 in its place. Old chats stay put.
OpenAI rolled out GPT-6 in ChatGPT on Oct. 7, the same day Haiku 5.5 arrived, along with an interface that drops interactive charts, calculators and buttons into answers. Product manager Aarush Selvan said the most helpful answers "aren't just text." The Haiku launch, meanwhile, capped a month in which Anthropic also moved Opus and Sonnet to version 5.5.
Read Next: Can USDC Win Over The Developing World? Circle Is About To Find Out
