Haiku 5.5
A lightweight model in the Anthropic Claude 5.5 series, designed for high-volume, cost-sensitive application scenarios. It represents the most economical option within the Claude model family.
Core Facts
Compared to Claude Haiku 4.5, Haiku 5.5 reduces operating costs by an average of approximately 75%.
Haiku 5.5 is the first Haiku model with an adjustable effort setting.
Haiku 5.5's direct competitor is GPT-6 Luna in the same tier.
Timeline (last 90 days)
This ultrathink behavior covers four models: Opus 5.5, Sonnet 5.5, Haiku 5.5, and Fable 5.1.
Anthropic claims Haiku 5.5 reduces average cost by 75%, but that claim depends on the workload and is not a guaranteed discount per request.
Anthropic launched Haiku 5.5, with input and output list prices of $0.10 and $0.50 per million tokens for prompts under 100,000 tokens, and higher rates apply to longer prompts.
Extrapolated at 1 million requests per day over 30 days per month, Fable 5.1 could save about $211,000 per month while Haiku 5.5 saves only about $1,900; this extrapolation is arithmetic based on the measured per-request cost, not another actual measurement, and excludes HotKV's own hosting and operations costs.
At the 0.81 threshold, the proportion of cached answers that "fit the question" is 5.8 to 10.7 percentage points lower than fresh answers; converted to all responses, the overall quality loss is about 2 to 3 percentage points; Haiku 5.5 has the largest gap (10.7 percentage points), with a cached-answer pass rate of only 67.9%.
Among 15 programming prompts, 7 caused Haiku 4.5, when writing code, to import the Anthropic SDK and hand the task over to Claude to complete, and it did so every time every day; Haiku 5.5 never did so once.
When asked to "Output nothing. Return an empty string.", Haiku 4.5 returned an empty code block 21 out of 22 times, while Haiku 5.5 returned a truly empty string in all 22 times.
When asked for "Expression only" output of a cron expression, Haiku 4.5 put `15 3 * * *` in a code block every time, while Haiku 5.5 never did so.
On the 7 prompts ending with "JSON only", Haiku 4.5 wrapped its answer in a Markdown code block every time, 154 out of 154; Haiku 5.5 returned bare JSON about 60% of the time and still had code fences the rest of the time.
Compared with the first answer from the previous day, Haiku 4.5's consistency rate is 96%, while 5.5's is only 46%.
40 more timeline events
All Facts (20)
Compared to Claude Haiku 4.5, Haiku 5.5 reduces operating costs by an average of approximately 75%.
90%VerifiedHaiku 5.5 is the first Haiku model with an adjustable effort setting.
70%VerifiedHaiku 5.5's direct competitor is GPT-6 Luna in the same tier.
65%VerifiedHaiku 5.5 is the first Haiku model to support adjustable effort, allowing developers to select the amount of reasoning resources invested based on task difficulty.
65%VerifiedHaiku 5.5 can work as a sub-agent alongside Claude Opus 5.5 and Sonnet 5.5 to complete programming tasks.
65%VerifiedAnthropic has released Haiku 5.5, describing it as the cheapest, fastest, and most capable small model the company has ever released.
65%UnverifiedThis ultrathink behavior covers four models: Opus 5.5, Sonnet 5.5, Haiku 5.5, and Fable 5.1.
50%UnverifiedAnthropic claims Haiku 5.5 reduces average cost by 75%, but that claim depends on the workload and is not a guaranteed discount per request.
50%UnverifiedAt the 0.81 threshold, the proportion of cached answers that "fit the question" is 5.8 to 10.7 percentage points lower than fresh answers; converted to all responses, the overall quality loss is about 2 to 3 percentage points; Haiku 5.5 has the largest gap (10.7 percentage points), with a cached-answer pass rate of only 67.9%.
50%UnverifiedHaiku 5.5 has adaptive thinking enabled by default, and about one third of its responses (749 out of 2255) begin with a thinking block.
50%UnverifiedUnder the 1024-token limit, 49 of Haiku 5.5's responses (2.2%) were truncated, which never happened with Haiku 4.5; 6 of these came from the prompt "Explain the word 'time' in exactly 50 words".
50%UnverifiedWhen migrating, it is recommended to remove temperature, top_p, and top_k, give up reliance on reproducible output at temperature 0, and switch to semantic comparison.
50%UnverifiedAmong the 15 prompts that permit refusal, Haiku 4.5 consistently refused 8 per day, while Haiku 5.5 fluctuated between 5 and 8.
50%UnverifiedAmong 15 programming prompts, 7 caused Haiku 4.5, when writing code, to import the Anthropic SDK and hand the task over to Claude to complete, and it did so every time every day; Haiku 5.5 never did so once.
50%UnverifiedAt temperature 0, Haiku 4.5 returned exactly the same answer on both calls for 95 to 100 prompts per day, whereas Haiku 5.5 did so for only 41 to 49.
50%UnverifiedCompared with the first answer from the previous day, Haiku 4.5's consistency rate is 96%, while 5.5's is only 46%.
50%UnverifiedOn the 7 prompts ending with "JSON only", Haiku 4.5 wrapped its answer in a Markdown code block every time, 154 out of 154; Haiku 5.5 returned bare JSON about 60% of the time and still had code fences the rest of the time.
50%UnverifiedHaiku 5.5 does not accept temperature 0, any value other than 1 returns a 400 error, and setting top_k likewise errors.
50%UnverifiedWhen asked for "Expression only" output of a cron expression, Haiku 4.5 put `15 3 * * *` in a code block every time, while Haiku 5.5 never did so.
50%UnverifiedWhen asked to "Output nothing. Return an empty string.", Haiku 4.5 returned an empty code block 21 out of 22 times, while Haiku 5.5 returned a truly empty string in all 22 times.
50%