name = "GPT-5.6 Terra"
description = "Balanced GPT-5.6 model for capable, cost-efficient everyday work"
family = "gpt-terra"
release_date = "2026-07-09"
last_updated = "2026-07-09"
attachment = true
reasoning = true
temperature = false
tool_call = true
structured_output = true
knowledge = "2026-02-16"
open_weights = false

[limit]
context = 1_050_000
input = 922_000
output = 128_000

[modalities]
input = ["text", "image", "pdf"]
output = ["text"]

[[benchmarks]]
name = "SWE-Bench Pro"
score = 63.4
metric = "resolve rate"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "Terminal-Bench"
score = 87.4
metric = "success rate"
version = "2.1"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "DeepSWE"
score = 69.6
metric = "resolve rate"
version = "1.1"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "GPQA Diamond"
score = 92.9
metric = "accuracy"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "FrontierMath"
score = 84.9
metric = "accuracy"
dataset = "Tier 1-3"
version = "v2"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "BrowseComp"
score = 87.5
metric = "accuracy"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "OSWorld"
score = 50.2
metric = "success rate"
version = "2.0"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "MMMU Pro"
score = 80.7
metric = "accuracy"
variant = "no tools"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "Agents' Last Exam"
score = 50.4
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "Toolathlon"
score = 53.1
metric = "success rate"
source = "https://openai.com/index/gpt-5-6/"
date = "2026-07-09"

[[benchmarks]]
name = "Artificial Analysis Intelligence Index"
score = 55
metric = "index score"
variant = "max"
version = "4.1"
source = "https://artificialanalysis.ai/articles/gpt-5-6-has-landed"
date = "2026-07-09"

[[benchmarks]]
name = "Artificial Analysis Coding Agent Index"
score = 77.4
metric = "index score"
harness = "Codex"
variant = "max"
version = "1.1"
source = "https://artificialanalysis.ai/articles/gpt-5-6-has-landed"
date = "2026-07-09"