Support

Settings

Theme

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

Mode

OpenAI · released 2026-10-26

GPT-6.1 Terra

Measured on 29 benchmarks across 6 reasoning efforts.

58.9

#22 of 125 models · best at max

includes estimates: 24% of the mix is measured for this model

Reasoning efforts in this ranking

max58.9#53 of 286
xhigh55.7#62 of 286
high54.9#65 of 286
medium50.1#81 of 286
low40.6#140 of 286
none30.1#212 of 286

Strengths

  • GPT-6.1 Terra
  • Opus 5.5 (#1)
  • Field median
bestField medianDeepSWE 1.1TB 4.0SimpleBenchHumanity's Last E…BSBenchARC-AGI-3FM Tier 4AA-Omniscience
  1. DeepSWE v1.117%65.6% est.#29 of 125
  2. Terminal-Bench 4.021%40.9% est.#23 of 125
  3. SimpleBench15%73.1% est.#27 of 125
  4. Humanity's Last Exam14%46.2%#31 of 125
  5. BullshitBench v25%61.9% est.#23 of 125
  6. ARC-AGI-33%56.4% est.#19 of 125
  7. FrontierMath Tier 4 (v2)16%69.8% est.#22 of 125
  8. AA-Omniscience9%29.8#21 of 125

Bars show where GPT-6.1 Terra sits among the 125 ranked models on each benchmark (right is best); the tick is the field median.

Cost and speed

against 125 ranked models

Cost per task

#15 cheapest of 73

$0.05at max

Output tokens per task

#26 fewest of 91

35kat max

Not measured for GPT-6.1 Terra: time per task, output speed, time to first token.

This ranking’s benchmarks

2 of 8 measured

reasoning

Humanity's Last Exam

601 results
  1. #46max of 60146.2%
  2. #57xhigh of 60143.7%
  3. #63high of 60143.3%
  4. #90medium of 60140.2%
  5. #119low of 60135.8%
  6. #215none of 60121.1%

Leader: Opus 5.5 · 61.4%

knowledge

AA-Omniscience

540 results
  1. #33max of 54029.8
  2. #39xhigh of 54028.7
  3. #42high of 54028.4
  4. #48medium of 54027.1
  5. #59low of 54023.7
  6. #120none of 5403.9

Leader: Opus 5.5 · 46.4

All 29 benchmarks

show 27 more

agentic

AA-Briefcase v1.1

238 results
  1. #58max of 2381402
  2. #71xhigh of 2381329
  3. #84high of 2381264
  4. #91none of 2381229
  5. #96medium of 2381192
  6. #124low of 2381030

Leader: Sonnet 5.5 · 1812

agentic

AutomationBench-AA

232 results
  1. #35max of 23260.2%
  2. #39xhigh of 23259.9%
  3. #46high of 23258.5%
  4. #51medium of 23257.8%
  5. #59low of 23256.2%
  6. #82none of 23251.0%

Leader: Gemini 4 · 77.5%

coding

FrontierCode 1.1-AA

77 results
  1. #19medium of 7743.7%
  2. #31xhigh of 7741.4%
  3. #32none of 7741.0%
  4. #43high of 7737.2%
  5. #47low of 7736.3%
  6. #48max of 7736.0%

Leader: Opus 5.5 · 57.1%

coding

SciCode

240 results
  1. #54max of 24055.1%
  2. #78xhigh of 24053.4%
  3. #88low of 24052.4%
  4. #96high of 24051.6%
  5. #98medium of 24051.6%
  6. #135none of 24047.8%

Leader: Opus 5.5 · 66.9%

coding

Terminal-Bench Science-AA

89 results
  1. #5max of 8957.6%
  2. #14xhigh of 8941.9%
  3. #19high of 8933.3%
  4. #23medium of 8924.8%
  5. #25none of 8924.3%
  6. #48low of 897.6%

Leader: GPT-6 Astra · 63.3%

coding

Terminal-Bench v4-AA

229 results
  1. #12max of 22955.6%
  2. #25xhigh of 22946.5%
  3. #33high of 22941.9%
  4. #47medium of 22932.3%
  5. #55none of 22926.8%
  6. #75low of 22915.2%

Leader: Kimi K3.1 · 67.2%

composite

AA Intelligence Index v4.3.2

624 results
  1. #26max of 62448.1
  2. #38xhigh of 62445.4
  3. #48high of 62443.9
  4. #62medium of 62441.3
  5. #90low of 62436.2
  6. #112none of 62433.5

Leader: Opus 5.5 · 57.6

cost

AA Intelligence Index Cost per Task

197 results
  1. #29low of 197$0.05
  2. #43medium of 197$0.10
  3. #47none of 197$0.11
  4. #59high of 197$0.13
  5. #74xhigh of 197$0.19
  6. #102max of 197$0.40

Leader: GPT-6 Luna · $0.00

cost

Cost to Run AA Intelligence Index

197 results
  1. #33low of 197$115
  2. #41none of 197$154
  3. #46medium of 197$175
  4. #57high of 197$239
  5. #79xhigh of 197$365
  6. #110max of 197$909

Leader: GPT-6 Luna · $10.63

creative

VoxelBench (text)

58 results
  1. #58max of 581000

Leader: GPT-6 Astra · 2744

economics

GDP.pdf

231 results
  1. #14max of 23128.6%
  2. #15medium of 23128.4%
  3. #17high of 23128.0%
  4. #26xhigh of 23126.6%
  5. #34low of 23125.6%
  6. #50none of 23123.2%

Leader: Muse Wonder Full · 32.6%

economics

GDPval-AA v2.1

304 results
  1. #57max of 3041468
  2. #76xhigh of 3041403
  3. #95high of 3041342
  4. #106medium of 3041316
  5. #111none of 3041282
  6. #126low of 3041199

Leader: Opus 5.5 · 1866

knowledge

AA-Omniscience Accuracy

540 results
  1. #62high of 54050.0%
  2. #66xhigh of 54049.6%
  3. #69medium of 54049.4%
  4. #70max of 54049.1%
  5. #79low of 54047.8%
  6. #101none of 54044.0%

Leader: Fable 5.1 · 67.2%

knowledge

AA-Omniscience Business

540 results
  1. #35max of 54023.3
  2. #42xhigh of 54022.0
  3. #45high of 54021.6
  4. #49medium of 54020.6
  5. #63low of 54015.2
  6. #97none of 5403.0

Leader: Fable 5 · 34.5

knowledge

AA-Omniscience Health

540 results
  1. #35max of 54014.0
  2. #41high of 54011.1
  3. #50xhigh of 5409.7
  4. #57medium of 5408.5
  5. #64low of 5406.3
  6. #150none of 540-12.2

Leader: GPT-6 Astra · 27.0

knowledge

AA-Omniscience Humanities & Social Sciences

540 results
  1. #31max of 54030.8
  2. #38xhigh of 54029.0
  3. #46high of 54027.0
  4. #49medium of 54026.1
  5. #65low of 54021.5
  6. #167none of 540-5.6

Leader: Opus 5.5 · 44.7

knowledge

AA-Omniscience Law

540 results
  1. #52xhigh of 54025.5
  2. #59max of 54024.6
  3. #61high of 54024.3
  4. #63medium of 54023.5
  5. #69low of 54020.1
  6. #183none of 540-10.7

Leader: Gemini 4 · 53.0

knowledge

AA-Omniscience Non-hallucination

540 results
  1. #72max of 54062.0%
  2. #85xhigh of 54058.5%
  3. #90high of 54056.9%
  4. #94medium of 54056.0%
  5. #107low of 54054.0%
  6. #234none of 54028.2%

Leader: MiniCPM5-1B (Non-reasoning) · 99.1%

knowledge

AA-Omniscience Science, Engineering & Math

540 results
  1. #64max of 54016.1
  2. #69xhigh of 54015.4
  3. #72medium of 54014.8
  4. #74high of 54014.6
  5. #96low of 54010.9
  6. #171none of 540-3.0

Leader: Opus 5.5 · 37.8

knowledge

AA-Omniscience Software Engineering

540 results
  1. #27high of 54071.7
  2. #36xhigh of 54070.6
  3. #37max of 54069.9
  4. #38medium of 54069.1
  5. #39low of 54068.3
  6. #69none of 54051.6

Leader: Opus 5.5 · 88.8

knowledge

MMMU-Pro

276 results
  1. #17max of 27684.7%
  2. #27xhigh of 27683.8%
  3. #28high of 27683.8%
  4. #34medium of 27682.8%
  5. #37low of 27682.5%
  6. #99none of 27676.0%

Leader: Opus 5.5 · 87.7%

long-context

AA-LCR v1.1

537 results
  1. #38max of 53782.7%
  2. #46high of 53782.3%
  3. #54xhigh of 53781.7%
  4. #66low of 53781.0%
  5. #72medium of 53780.7%
  6. #244none of 53767.7%

Leader: Kimi K3 · 88.7%

reasoning

CritPt

543 results
  1. #28max of 54328.0%
  2. #39xhigh of 54326.6%
  3. #43high of 54325.7%
  4. #61medium of 54321.4%
  5. #97low of 54315.1%
  6. #165none of 5434.9%

Leader: Muse Wonder Full · 32.9%

reliability

Tool-call error rate

177 results
  1. #11max of 1770.1%

Leader: GPT-6 Sol Pro · 0.0%

reliability

Uptime (7d)

186 results
  1. #6max of 186100.0%

Leader: GPT-5 Pro · 100.0%

speed

Provider latency

181 results
  1. #154max of 1813.2s

Leader: Llama 3.2 1b · 0.2s

speed

Provider throughput

181 results
  1. #44max of 18189.8

Leader: Muse Spark 1.2 · 241.0