Skip to content

GPT-5.6 Terra

OpenAI · Released 9 Jul 2026 · gpt-5.6-terra

Compare this model
ProprietaryCodingBudget

$2.00

Input price / 1M tokens

#36 of 61, cheapest first

$12.00

Output price / 1M tokens

#44 of 61, cheapest first

1.05M

Context window

#2 of 71

97

Output tokens / second

#30 of 61

63.4%

SWE-bench Pro

#3 of 16

69.6%

DeepSWE v1.1

#8 of 17

87.4%

Terminal-Bench 2.1

#7 of 17

50.2%

OSWorld 2.0

#8 of 10

Summary

GPT-5.6 Terra is a proprietary model from OpenAI, released on 9 Jul 2026.

At $2.00 input and $12.00 output per million tokens, it is #36 of 61 on input price, cheapest first.

Its 1.05M-token context window ranks #2 of 71.

Measured output speed is 97 tokens per second, #30 of 61.

The balanced GPT-5.6 tier for workloads that balance intelligence and cost. It is the successor to the earlier 'mini' tier.

Benchmarks · 5 reported

BenchmarkScoreBarRankSource
SWE-bench Pro63.4%#3 of 16OpenAI launch postSelf-reported
DeepSWE v1.169.6%#8 of 17OpenAI launch postSelf-reported
Terminal-Bench 2.187.4%#7 of 17OpenAI launch postSelf-reported
OSWorld 2.050.2%#8 of 10OpenAI launch postSelf-reported
GPQA Diamond92.9%#6 of 27OpenAI launch postSelf-reported

Self-reported means the lab ran the test itself. A rank appears only where several labs report the same version of a benchmark.

Similar models

Models with the most benchmarks in common with GPT-5.6 Terra, and the closest scores.

Compare 4 side by side

SWE-bench Pro

#3 of 16

Harder multi-file engineering tasks

DeepSWE v1.1

#8 of 17

Long coding tasks in real repos

Terminal-Bench 2.1

#7 of 17

Terminal tasks, earlier set

OSWorld 2.0

#8 of 10

Long desktop and web workflows

Compare GPT-5.6 Terra side by side

Pre-filled with the two closest models. Swap any of them.

Open comparison

Sources

Building on GPT-5.6 Terra?

Get the architecture right before the bill arrives.

We size caching, routing and fallbacks for your workload on a free call, and tell you where a cheaper model is good enough.