priorsBuilder
Support

Settings

Theme

Explore themes

Mode

Library

Looks borrowed from benchmark sites and products. Themes with a light and a dark version follow your mode.

Bring your own theme

Upload a Priors theme file, or have your agent make one: copy the prompt, tell it the look you want, and upload what it returns. With the Priors skill installed it already knows the format.

Custom themes stay in this browser. They travel inside shared chart links and a page's theme, so others see them as you do.

For the builder, Explore and model pages on this device. Published pages keep the theme their owner picked.

Mode

Anthropic · released 2026-10-07

Haiku 5.5

Measured on 145 benchmarks across 5 reasoning efforts.

36.2

#16 of 61 models · best at max

Reasoning efforts in this ranking

max36.2#39 of 139
xhigh34.3#47 of 139
high29.6#65 of 139
medium24.5#85 of 139
low18.6#103 of 139

Strengths

  • Haiku 5.5
  • Gemini 4 Carbon (#1)
  • Field median
bestField medianFrontierCode 2.0-…Terminal-Bench v4…AA-Briefcase v1.1AutomationBench-AATerminal-Bench-Sc…HLE-DiamondCritPt v2 Hard Su…AA-Omniscience v1…AA-Omniscience v1…GDP.pdf
  1. FrontierCode 2.0-AA15%34.9%#9 of 61
  2. Terminal-Bench v4-AA10%32.8%#22 of 61
  3. AA-Briefcase v1.115%1576#12 of 61
  4. AutomationBench-AA5%35.4%#42 of 61
  5. Terminal-Bench-Science-AA15%20.0%#15 of 61
  6. HLE-Diamond10%29.6%#17 of 61
  7. CritPt v2 Hard Subset10%41.0%#18 of 61
  8. AA-Omniscience v1.1 Accuracy10%34.7%#34 of 61
  9. AA-Omniscience v1.1 Non-hallucination5%65.8%#21 of 61
  10. GDP.pdf5%20.8%#22 of 61

Bars show where Haiku 5.5 sits among the 61 ranked models on each benchmark (right is best); the tick is the field median.

Cost and speed

against 61 ranked models

Cost per task

#10 cheapest of 41

$0.009at max

Time per task

#17 quickest of 40

9.9 minat max

Output speed

#6 fastest of 52

235 tok/sat max

Time to first token

#49 quickest of 52

5.6 minat max

Output tokens per task

#43 fewest of 44

226kat max

Token price

$0.018 per 1M tokens at API prices$0.00037 on Claude Max 20x

95.8% cache read · 0.28% fresh input · 3.13% cache write · 0.79% output

Measured from T3 Code usage in Claude Code, Oct 27 to Nov 2, high confidence. One app’s users, not everyone’s. $0.015 at the standard 97 / 2.5 / 0.5 mix.

This ranking’s benchmarks

10 of 10 measured

agentic

AutomationBench-AA

241 results
  1. #121xhigh of 24136.0%
  2. #124max of 24135.4%
  3. #128high of 24133.7%
  4. #133medium of 24128.6%
  5. #146low of 24122.9%

Leader: Gemini 4 Carbon · 77.8%

agentic

AA-Briefcase v1.1

247 results
  1. #22max of 2471576
  2. #34xhigh of 2471525
  3. #53high of 2471441
  4. #68medium of 2471369
  5. #116low of 2471114

Leader: Sonnet 5.5 · 1814

reasoning

CritPt v2 Hard Subset

100 results
  1. #37xhigh of 10044.5%
  2. #40high of 10042.5%
  3. #42max of 10041.0%
  4. #57medium of 10030.0%
  5. #73low of 10021.5%

Leader: GPT-6 Astra · 71.5%

coding

FrontierCode 2.0-AA

97 results
  1. #20max of 9734.9%
  2. #27xhigh of 9729.9%
  3. #42high of 9723.2%
  4. #59medium of 9716.1%
  5. #75low of 978.0%

Leader: Opus 5.5 · 46.4%

economics

GDP.pdf

240 results
  1. #68max of 24020.8%
  2. #86xhigh of 24018.2%
  3. #95high of 24017.2%
  4. #109medium of 24015.2%
  5. #136low of 24011.2%

Leader: Muse Wonder Full · 32.6%

reasoning

HLE-Diamond

142 results
  1. #40max of 14229.6%
  2. #42xhigh of 14228.8%
  3. #54high of 14223.7%
  4. #72medium of 14216.8%
  5. #84low of 14214.1%

Leader: Gemini 4 Carbon · 69.1%

knowledge

AA-Omniscience v1.1 Accuracy

111 results
  1. #69max of 11134.7%
  2. #70xhigh of 11133.8%
  3. #74high of 11131.8%
  4. #76medium of 11131.7%
  5. #79low of 11130.4%

Leader: Fable 5.1 · 68.4%

knowledge

AA-Omniscience v1.1 Non-hallucination

111 results
  1. #29max of 11165.8%
  2. #40high of 11161.1%
  3. #41xhigh of 11160.8%
  4. #43medium of 11159.6%
  5. #44low of 11159.5%

Leader: Gemini 4 Carbon · 95.8%

reasoning

Terminal-Bench-Science-AA

115 results
  1. #38max of 11520.0%
  2. #39xhigh of 11518.1%
  3. #53high of 11510.0%
  4. #62medium of 1157.1%
  5. #84low of 1151.9%

Leader: Gemini 4 Carbon · 64.8%

coding

Terminal-Bench v4-AA

238 results
  1. #48max of 23832.8%
  2. #56xhigh of 23829.3%
  3. #71high of 23821.7%
  4. #84medium of 23815.2%
  5. #94low of 23812.6%

Leader: Kimi K3.1 · 67.2%

All 145 benchmarks

show 135 more