DataLearner logo
GP

GPT-5.6 Terra

Reasoning modelCoding modelGPT-5.6

GPT-5.6 Terra

Release date: 2026-06-26Updated: 2026-08-23Knowledge cutoff: 2026-02-16Views: 670
Live demoGitHubHugging FaceCompare
Parameters
No data
Context length
1.05M
Multilingual
Supported
Reasoning ability
4/5

GPT-5.6 Terra is an AI model from OpenAI, released on 2026-06-26.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

GPT-5.6 Terra

Model basics

Reasoning traces
Supported
Thinking modes
Thinking Level · Medium (Default)Standard ModeThinking Level · LowThinking Level · High
Context length
1.05M tokens
Max output length
128K tokens
Model type
Reasoning model
Modality (in / out)
Text, Image → Text
Release date
2026-06-26
Model file size
No data
MoE architecture
No
Total params / Active params
No data / Not applicable
Knowledge cutoff
2026-02-16
GPT-5.6 Terra

Open source & experience

Code license
Proprietary
Weights license
Proprietary
GitHub repo
N/A
Hugging Face
N/A
Live demo
N/A
GPT-5.6 Terra

Official resources

GPT-5.6 Terra

API details

API speed
4/5
💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.
Standard
TypeConditionInputOutput
Text-$2.00/ 1M$12.00/ 1M
long-context
TypeConditionInputOutput
Text-$4.00/ 1M$18.00/ 1M
Cache PricingPrompt Cache
TypeTTLWriteRead
Text30m$2.50/ 1M$0.200/ 1M

“—” means the modality is not billed in that direction, or the vendor has not published a price for it.

GPT-5.6 Terra

Benchmark Results

GPT-5.6 Terra currently shows benchmark results led by FrontierMath v2 (4 / 58, score 85.96), GPQA Diamond (19 / 274, score 93.31), Context Arena (11 / 126, score 92.17). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
Tool usage

General Knowledge

5 evaluations
Benchmark / mode
Score
Rank/total
96.50
9 / 92
83.90
19 / 85
Vals Index
Extra-High
65.14
2 / 2

General Evaluation

3 evaluations
Benchmark / mode
Score
Rank/total
GPQA Diamond
Standard Mode
77.27
164 / 274
87.37
81 / 274
93.31
19 / 274

Writing and Creative Capabilities

1 evaluations
Benchmark / mode
Score
Rank/total
Creative Writing
Standard Mode
1850.30
11 / 106

Common Sense Reasoning

1 evaluations
Benchmark / mode
Score
Rank/total
SimpleBench
Extra-High
48.90
50 / 92

Text Embedding

1 evaluations
Benchmark / mode
Score
Rank/total
92.17
11 / 126

Long Context

2 evaluations
Benchmark / mode
Score
Rank/total
79.67
3 / 29
AA-LCR
Extra-High
75
7 / 29

AI Agent - Tool Usage

4 evaluations
Benchmark / mode
Score
Rank/total
87.40
12 / 53
78.40
29 / 53
21.52
13 / 20
8.60
7 / 11

Coding and Software Engineer

6 evaluations
Benchmark / mode
Score
Rank/total
1521.61
16 / 35
WeirdML v2
HighTools
78.27
10 / 52
DeepSWE
Extra-HighTools
69.60
9 / 38
53.94
7 / 16
SciCode
Extra-High
51.62
9 / 16

Agent Level Benchmark

3 evaluations
Benchmark / mode
Score
Rank/total
Agents' Last Exam
Extra-HighTools
50.40
4 / 19
τ³-Banking
MaxTools
40.21
2 / 13
τ³-Banking
Extra-HighTools
29.69
7 / 13

Productivity Knowledge

3 evaluations
Benchmark / mode
Score
Rank/total
GDPval-AA v2
MaxTools
1565.62
13 / 27
GDPval-AA v2
Extra-HighTools
1562.99
14 / 27
AA-Briefcase
MaxTools
1349.73
12 / 20

Math and Reasoning

2 evaluations
Benchmark / mode
Score
Rank/total
85.96
4 / 58

Compare with other models

GPT-5.6 Terra

Publisher

GPT-5.6 Terra

Model Overview

GPT-5.6 Terra

GPT-5.6 Terra is OpenAI's balanced mid-tier model in the GPT-5.6 family, first previewed on June 26, 2026 and made generally available on July 9, 2026. It targets high-concurrency production workloads — customer support, internal tooling, document analysis — delivering roughly GPT-5.5-level capability at about half the price. The family also includes Sol (the flagship, for the hardest coding/research/security tasks) and Luna (the most cost-efficient tier).

On Artificial Analysis's Intelligence Index (max reasoning), Terra scores 55 versus Sol's 59, at roughly half the cost per task (~$0.55 vs ~$1.04). It reaches 87.40 on Terminal-Bench 2.1, 69.60 on DeepSWE, 77.4 on the AA Coding Agent Index, 50.4 on Agents' Last Exam, and 65.14% accuracy on the Vals Index. Independent testing by CodeRabbit found Terra completes 40.7% of long-horizon coding tasks (versus Sol's 63.7%) while producing notably more verbose output per task — best suited to well-scoped, cost-sensitive coding work rather than long or highly complex jobs.

Terra shares OpenAI's GPT-5.6 system card with Sol and Luna, including "High" Preparedness Framework ratings for cybersecurity and biological/chemical risk; see the GPT-5.6 Sol page for the detailed safety-evaluation breakdown.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code