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

Google · released 2026-10-28

Gemini 4 Carbon

Measured on 62 benchmarks across 3 reasoning efforts.

59.4

#1 of 61 models · best at high

Reasoning efforts in this ranking

high59.4#1 of 139
medium57.7#3 of 139
low53.7#10 of 139

Strengths

  • Gemini 4 Carbon
  • 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%40.6%#6 of 61
  2. Terminal-Bench v4-AA10%65.2%#2 of 61
  3. AA-Briefcase v1.115%1610#9 of 61
  4. AutomationBench-AA5%77.8%#1 of 61
  5. Terminal-Bench-Science-AA15%64.8%#1 of 61
  6. HLE-Diamond10%69.1%#1 of 61
  7. CritPt v2 Hard Subset10%63.0%#4 of 61
  8. AA-Omniscience v1.1 Accuracy10%58.3%#6 of 61
  9. AA-Omniscience v1.1 Non-hallucination5%88.6%#2 of 61
  10. GDP.pdf5%28.4%#7 of 61

Bars show where Gemini 4 Carbon 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

#39 cheapest of 41

$2.8at high

Time per task

#33 quickest of 40

21 minat high

Output speed

#32 fastest of 52

94 tok/sat high

Time to first token

#45 quickest of 52

2.5 minat high

Output tokens per task

#34 fewest of 44

123kat high

Token price

$0.850 per 1M tokens at API prices$0.303 on Google AI Pro (assumed same as Argon)

85.0% cache read · 14.5% fresh input · 0.00% cache write · 0.50% output

Assumed mix: no measured T3 Code usage for this model.

This ranking’s benchmarks

10 of 10 measured

agentic

AutomationBench-AA

241 results
  1. #1high of 24177.8%
  2. #2medium of 24177.6%
  3. #28low of 24163.5%

Leader: this model · 77.8%

agentic

AA-Briefcase v1.1

247 results
  1. #17high of 2471610
  2. #19medium of 2471602
  3. #26low of 2471568

Leader: Sonnet 5.5 · 1814

reasoning

CritPt v2 Hard Subset

100 results
  1. #10medium of 10063.5%
  2. #11high of 10063.0%
  3. #16low of 10060.0%

Leader: GPT-6 Astra · 71.5%

coding

FrontierCode 2.0-AA

97 results
  1. #8high of 9740.6%
  2. #10medium of 9739.5%
  3. #28low of 9729.6%

Leader: Opus 5.5 · 46.4%

economics

GDP.pdf

240 results
  1. #13medium of 24029.6%
  2. #15low of 24028.6%
  3. #17high of 24028.4%

Leader: Muse Wonder Full · 32.6%

reasoning

HLE-Diamond

142 results
  1. #1high of 14269.1%
  2. #3medium of 14267.4%
  3. #6low of 14261.9%

Leader: this model · 69.1%

knowledge

AA-Omniscience v1.1 Accuracy

111 results
  1. #21high of 11158.3%
  2. #27low of 11149.2%
  3. #28medium of 11148.9%

Leader: Fable 5.1 · 68.4%

knowledge

AA-Omniscience v1.1 Non-hallucination

111 results
  1. #1medium of 11195.8%
  2. #3high of 11188.6%
  3. #8low of 11180.2%

Leader: this model · 95.8%

reasoning

Terminal-Bench-Science-AA

115 results
  1. #1high of 11564.8%
  2. #3medium of 11562.4%
  3. #5low of 11561.9%

Leader: this model · 64.8%

coding

Terminal-Bench v4-AA

238 results
  1. #2high of 23865.2%
  2. #4low of 23862.6%
  3. #5medium of 23860.6%

Leader: Kimi K3.1 · 67.2%

All 62 benchmarks

show 52 more