OpenAI

GPT-5.6 Terra

Tool callingStructured outputReasoningImage inputFile input

Balanced GPT-5.6 model for capable, cost-efficient everyday work

Context
1.1M
Released
3mo ago
Input / 1M tokens
$2.00
Output / 1M tokens
$12.00
Cached input / 1M tokens
$0.20
Cache write / 1M tokens
$2.50

Sources:models.devOpenRouter

Benchmarks

DeepSWE is the primary coding benchmark where an exact model evaluation exists. Other sourced evaluations appear below; missing scores are not zero.

DeepSWE 1.1

Primary coding-agent benchmark · 113 tasks · Data generated 2026-09-22

Official leaderboard
69.6%
pass@1 · highest published configuration
mini-swe-agent · max reasoning
pass@4
88.5%
Average cost / attempt
$4.95
Benchmark runs
4

All published configurations

Harness / reasoningpass@1pass@4Cost / attempt
mini-swe-agent · max69.6%88.5%$4.95
mini-swe-agent · xhigh60.2%80.5%$2.13
mini-swe-agent · high53.8%80.5%$1.13
mini-swe-agent · medium35.1%60.2%$0.58
mini-swe-agent · low24.1%44.2%$0.43

pass@1 is the success rate across scored attempts; pass@4 is the share of tasks solved in at least one of four attempts. Costs are reported by DeepSWE for each configuration. These results are not directly comparable to evaluations using different harnesses. Official source JSON.

Other published benchmarks

EvaluationReported score and source
ARC-AGI-2
67.1%
Third-party evaluator · gpt-5.6-terra_high
Reasoning effort: high
External evaluation via Epoch AI
Original-source terms apply
ARC-AGI-2
18.8%
Third-party evaluator · gpt-5.6-terra_low
Reasoning effort: low
External evaluation via Epoch AI
Original-source terms apply
ARC-AGI-2
83.9%
Third-party evaluator · gpt-5.6-terra_max
Reasoning effort: max
External evaluation via Epoch AI
Original-source terms apply
ARC-AGI-2
37.5%
Third-party evaluator · gpt-5.6-terra_medium
Reasoning effort: medium
External evaluation via Epoch AI
Original-source terms apply
ARC-AGI-2
74.2%
Third-party evaluator · gpt-5.6-terra_xhigh
Reasoning effort: xhigh
External evaluation via Epoch AI
Original-source terms apply
FrontierMath-Tier-4-v2-Private
70.7%
Epoch AI · gpt-5.6-terra_max
Reasoning effort: max
Epoch AI evaluation
FrontierMath-Tiers-1-3-v2-Private
86.0%
Epoch AI · gpt-5.6-terra_max
Reasoning effort: max
Epoch AI evaluation
GPQA diamond
87.4%
Epoch AI · gpt-5.6-terra_low
Reasoning effort: low
Epoch AI evaluation
GPQA diamond
93.3%
Epoch AI · gpt-5.6-terra_max
Reasoning effort: max
Epoch AI evaluation
GPQA diamond
77.3%
Epoch AI · gpt-5.6-terra_none
Reasoning effort: none
Epoch AI evaluation
OTIS Mock AIME 2024-2025
88.9%
Epoch AI · gpt-5.6-terra_low
Reasoning effort: low
Epoch AI evaluation
OTIS Mock AIME 2024-2025
99.7%
Epoch AI · gpt-5.6-terra_max
Reasoning effort: max
Epoch AI evaluation
OTIS Mock AIME 2024-2025
53.3%
Epoch AI · gpt-5.6-terra_none
Reasoning effort: none
Epoch AI evaluation
SimpleQA Verified
43.2%
Epoch AI · gpt-5.6-terra_max
Reasoning effort: max
Epoch AI evaluation

Benchmark compilation: Epoch AI, Capabilities & benchmarking (CC BY attribution). Scores are converted to percentages and model names matched to the Vertical Tabs catalog. Individual evaluations are credited to their original sources. External results may carry separate licences and evaluation configurations; do not interpret scores from different harnesses as directly comparable.

Specifications

Developer
OpenAI
Model family
GPT
Type
Language model
Predecessor
—
Successor
—
Released
July 9, 2026
Model ID
openai/gpt-5.6-terra
Openness
Closed
License
—
Input modalities
file, image, text
Output modalities
text
Modality
text+image+file->text
Tokenizer
GPT
Max output tokens
128K
Knowledge cutoff
2026-02-16
Hugging Face
—
Source freshness
Metadata synced Oct 8, 2026, 10:17 PM UTC
Additional source evidence
Source model
openai/gpt-5.6-terra
Source family
gpt-terra
Last updated
2026-07-09
Knowledge cutoff
2026-02-16
Model context
1.1M
Maximum input
922K
Model output limit
128K
Open weights
No
File attachments
Yes
Reasoning
Yes
Tool calling
Yes
Structured output
Yes
Temperature control
No

Published benchmarks

BenchmarkScoreEvaluation detailsEvidence
SWE-Bench Pro63.4 · resolve rate2026-07-09Source
Terminal-Bench87.4 · success rate2.1 · 2026-07-09Source
DeepSWE69.6 · resolve rate1.1 · 2026-07-09Source
GPQA Diamond92.9 · accuracy2026-07-09Source
FrontierMath84.9 · accuracyTier 1-3 · v2 · 2026-07-09Source
BrowseComp87.5 · accuracy2026-07-09Source
OSWorld50.2 · success rate2.0 · 2026-07-09Source
MMMU Pro80.7 · accuracyno tools · 2026-07-09Source
Agents' Last Exam50.42026-07-09Source
Toolathlon53.1 · success rate2026-07-09Source
Artificial Analysis Intelligence Index55 · index scoremax · 4.1 · 2026-07-09Source
Artificial Analysis Coding Agent Index77.4 · index scoreCodex · max · 1.1 · 2026-07-09Source
Complete source record (JSON)
{
  "id": "openai/gpt-5.6-terra",
  "name": "GPT-5.6 Terra",
  "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
  "family": "gpt-terra",
  "last_updated": "2026-07-09",
  "knowledge": "2026-02-16",
  "attachment": true,
  "reasoning": true,
  "tool_call": true,
  "structured_output": true,
  "temperature": false,
  "modalities": {
    "input": [
      "text",
      "image",
      "pdf"
    ],
    "output": [
      "text"
    ]
  },
  "limit": {
    "context": 1050000,
    "input": 922000,
    "output": 128000
  },
  "benchmarks": [
    {
      "name": "SWE-Bench Pro",
      "score": 63.4,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "resolve rate",
      "date": "2026-07-09"
    },
    {
      "name": "Terminal-Bench",
      "score": 87.4,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "success rate",
      "version": "2.1",
      "date": "2026-07-09"
    },
    {
      "name": "DeepSWE",
      "score": 69.6,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "resolve rate",
      "version": "1.1",
      "date": "2026-07-09"
    },
    {
      "name": "GPQA Diamond",
      "score": 92.9,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "accuracy",
      "date": "2026-07-09"
    },
    {
      "name": "FrontierMath",
      "score": 84.9,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "accuracy",
      "dataset": "Tier 1-3",
      "version": "v2",
      "date": "2026-07-09"
    },
    {
      "name": "BrowseComp",
      "score": 87.5,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "accuracy",
      "date": "2026-07-09"
    },
    {
      "name": "OSWorld",
      "score": 50.2,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "success rate",
      "version": "2.0",
      "date": "2026-07-09"
    },
    {
      "name": "MMMU Pro",
      "score": 80.7,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "accuracy",
      "variant": "no tools",
      "date": "2026-07-09"
    },
    {
      "name": "Agents' Last Exam",
      "score": 50.4,
      "source": "https://openai.com/index/gpt-5-6/",
      "date": "2026-07-09"
    },
    {
      "name": "Toolathlon",
      "score": 53.1,
      "source": "https://openai.com/index/gpt-5-6/",
      "metric": "success rate",
      "date": "2026-07-09"
    },
    {
      "name": "Artificial Analysis Intelligence Index",
      "score": 55,
      "source": "https://artificialanalysis.ai/articles/gpt-5-6-has-landed",
      "metric": "index score",
      "variant": "max",
      "version": "4.1",
      "date": "2026-07-09"
    },
    {
      "name": "Artificial Analysis Coding Agent Index",
      "score": 77.4,
      "source": "https://artificialanalysis.ai/articles/gpt-5-6-has-landed",
      "metric": "index score",
      "harness": "Codex",
      "variant": "max",
      "version": "1.1",
      "date": "2026-07-09"
    }
  ],
  "release_date": "2026-07-09",
  "open_weights": false
}

Infrastructure

Provider availability

All providers
ProviderEndpointContextInput /M tokensOutput /M tokensCached input /M tokensCache write /M tokensUptime 1dData handling
OpenAIopenai/flex1.1M$1.00$6.00$0.10$1.2599.39%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

OpenAIopenai/flexBatch pricing1.1M$1.00$6.00$0.10$1.2599.43%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure1.1M$2.00$12.00$0.20$2.50100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

AzureazureBatch pricing1.1M$2.00$12.00$0.20$2.50100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

OpenAIopenai1.1M$2.00$12.00$0.20$2.50100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

OpenAIopenaiBatch pricing1.1M$2.00$12.00$0.20$2.50100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Amazon Bedrockamazon-bedrock/us-east-11.1M$2.20$13.20$0.22$2.75100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Amazon Bedrockamazon-bedrock/us-east-1Batch pricing1.1M$2.20$13.20$0.22$2.75100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure/us1.1M$2.20$13.20$0.22$2.75100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure/eu1.1M$2.20$13.20$0.22$2.75100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure/usBatch pricing1.1M$2.20$13.20$0.22$2.75100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure/euBatch pricing1.1M$2.20$13.20$0.22$2.75100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

OpenAIopenai/fast1.1M$4.00$24.00$0.40$5.00100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

OpenAIopenai/fastBatch pricing1.1M$4.00$24.00$0.40$5.00100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Provider data:OpenRouter

Text provenance

EU · product-scoped

OpenAI textGrain watermarking

In the EU, ChatGPT-generated text uses textGrain. OpenAI says API customers can opt in globally for select models, so watermarking depends on the product surface and configuration.

Official explanation