OpenAI
GPT-5.6 Terra Pro
Released 
July 2026

OpenAI
GPT-5.6 Terra Pro

GPT-5.6 Terra served in pro reasoning mode for higher-accuracy coding, agentic, and multi-step reasoning tasks that tolerate latency.

Modality:
Text
Image
PDF
model ID
openai/gpt-5.6-terra-pro

Output Speed *

N/A
tok/s

Intelligence Index *

N/A
/ 100

Context Window *

1050000
tokens

Input price

12
Anytoken

Output price

72
Anytoken
GPT-5.6 Terra Pro: Balanced-Tier Reasoning at Pro Mode Depth GPT-5.6 Terra Pro is OpenAI's mid-tier GPT-5.6 model served with reasoning.mode set to pro. It is the same underlying model as GPT-5.6 Terra, which sits between the flagship Sol tier and the cost-efficient Luna tier, but pro mode performs more internal work per request for higher-quality answers on complex tasks. It accepts text, image, and PDF input and returns text, with a 1,050,000-token context window. It fits coding, agentic orchestration, and professional reasoning workloads where accuracy matters more than raw response speed. Access GPT-5.6 Terra Pro through the AnyAPI.ai API

Performance

Where Pro Mode Earns Its Latency

GPT-5.6 Terra Pro targets complex, multi-step problems where an extra round of internal reasoning improves accuracy. OpenAI documents that pro mode performs more model work per request than the standard default, aggregating that work into a single answer at higher token usage and latency. On the underlying Terra model, Artificial Analysis measured 52 on its Intelligence Index at xhigh effort, above the reasoning-model median. That capability headroom matters for coding agents, planning, and analytical review, where a correct first pass avoids expensive retries. In production, the trade-off is deliberate: you accept slower, costlier responses for stronger outcomes on hard tasks.

Benchmarks

What Independent Testing Shows for the Terra Line

Independent measurement is available for the underlying GPT-5.6 Terra model rather than the pro-mode variant specifically. Artificial Analysis scored Terra at 52 on its Intelligence Index at xhigh effort and 42 at max effort, both above the reasoning-model median. Terra (max) reached 77 on the Coding Agent Index, with substantial per-task cost reductions versus flagship Sol. Speed and token usage scale sharply with effort: Terra (high) generated roughly 87 tokens per second with a ~4.8s time to first token. Treat these as directional evidence for Terra Pro rather than exact pro-mode figures.

Output Speed

*
N/A
tok/s

Intelligence Index

*
N/A
/ 100

MMLU *

Broad world knowledge and problem-solving
0
%

GPQA *

PhD-level scientific reasoning across physics, biology, chemistry.
0
%

HLE *

Adherence to multi-step structured instructions.
0
%

LiveCodeBench *

Tool-calling reliability in long agentic loops.
0
%

Technical Specifications

What the model supports

GPT-5.6 Terra Pro accepts text, image, and PDF input and returns text only; it is not an image or audio generator. The 1,050,000-token context window supports large repositories, document sets, and long histories in a single request, but prompts above 272K input tokens enter a higher long-context pricing tier and a large window does not guarantee perfect recall. Maximum output is 128,000 tokens per completion. Pro mode runs through reasoning.mode=pro and works best via the Responses API, which supports function calling, structured outputs, and persisted reasoning.
Verified Specifications — 
GPT-5.6 Terra Pro
*
Input modalities
Text
Image
PDF
output modalities
Text
Context window
1050000
 tokens
Maximum output tokens
128000
Reasoning
Yes
Knowledge cutoff
July 2026
Pricing (standard)
12
 AnyTokens in
 / 
72
 AnyTokens out

Comparison

GPT-5.6 Terra Pro vs GPT-5.6 Terra: Standard or Pro Mode?

GPT-5.6 Terra Pro is the same underlying model as GPT-5.6 Terra, differing only in reasoning.mode. Terra runs standard mode by default; Terra Pro runs pro mode, which performs more internal work per request for higher-quality answers on complex tasks. Both share the 1,050,000-token context window, 128,000-token maximum output, text/image/PDF input, tool calling, and structured outputs. The practical decision is not a new integration but a mode setting: how much extra latency and token usage you are willing to spend to raise answer quality on your hardest requests.

Dimension
GPT-5.6 Terra Pro
GPT-5.6 Terra (max)
Context window *
1050000
tokens
1050000
tokens
Output speed *
N/A
tok/s
119.18
tok/s
Intelligence Index *
N/A
42.3
Input pricing
12
AnyToken
12
AnyToken
Output pricing
72
AnyToken
72
AnyToken
Knowledge cutoff *
July 2026
July 2026

Choose GPT-5.6 Terra Pro when tasks are genuinely hard — long-horizon agents, intricate debugging, or analytical review — and a more accurate first pass justifies slower, costlier responses. Choose standard GPT-5.6 Terra when latency and throughput matter more, for everyday coding, extraction, and interactive workloads where standard mode already meets your quality bar. Because they share one model slug family, you can route only your most demanding requests to pro mode and keep the rest on standard, controlling cost without a second integration.

Limitations & Trade-offs

Where GPT-5.6 Terra Pro falls short

1
Higher latency and token usage by design. Pro mode performs more internal reasoning per request and, as OpenAI documents, bills that additional work at standard rates, so expect higher token consumption and slower responses than standard Terra. This makes Terra Pro a poor fit for latency-sensitive or interactive UIs. For chat surfaces and high-throughput pipelines, standard GPT-5.6 Terra or the lower-latency Luna tier are better defaults.
2
Text-only output. GPT-5.6 Terra Pro accepts text, image, and PDF input but returns text only. It does not generate images, audio, or video. Applications needing generated visuals or speech must pair it with a dedicated generation model rather than relying on Terra Pro's output modality.
3
Long-context pricing and recall limits. The 1,050,000-token window is large, but prompts above 272K input tokens enter a higher long-context pricing tier, and a wide context does not guarantee perfect fact retrieval across it. For very large repositories or document sets, targeted retrieval often beats submitting everything, both for cost control and for accuracy on specific facts.
4
Independent benchmarks are for Terra, not Terra Pro. Public Artificial Analysis scores cover the underlying GPT-5.6 Terra across effort levels rather than the pro-mode variant specifically. Treat those numbers as directional evidence for Terra Pro's ceiling and validate pro mode on your own workload before committing to it in production.

Best-Fit Workloads

Where this model earns its place

01

Complex coding and refactoring agents


Terra Pro's extra reasoning suits multi-file changes, debugging, patch preparation, and review where a correct first pass avoids retry loops. The underlying Terra model scored 77 on the Artificial Analysis Coding Agent Index at max effort with strong per-task cost reductions versus Sol. Route only hard coding tasks to pro mode and keep routine edits on standard Terra to manage latency and cost.

02

Long-horizon agentic orchestration


With tool calling, structured outputs, and Responses API support for persisted reasoning, Terra Pro fits agents that plan, call tools, and process intermediate results across multiple steps. Pro mode's additional internal work helps recover from ambiguity and choose better next actions. It is best reserved for the difficult decision points in an agent loop rather than every step, given its higher latency.

03

Analytical review over long documents


The 1,050,000-token window lets Terra Pro ingest large contracts, technical docs, logs, or reports in one request, while pro mode improves reasoning quality on the analysis. This suits failure-mode review, compliance checks, and technical due diligence. Watch the long-context pricing tier above 272K input tokens and prefer targeted retrieval when only specific facts are needed.

04

High-stakes structured extraction


For extraction where errors are expensive — regulatory data, financial figures, or clinical fields — Terra Pro's JSON schema support plus pro-mode reasoning improves accuracy on ambiguous or messy source material. Use it for the hardest cases and fall back to standard Terra or a cheaper model for clean, high-volume documents where speed matters more.

Pricing in anytokens via AnyAPI
Input
12
Output
72
Cache write
15
Cache read
1.2

Integration

Access GPT-5.6 Terra Pro via AnyAPI.ai

Access GPT-5.6 Terra Pro through AnyAPI.ai using a unified API built for multi-model AI applications. Integrate GPT-5.6 Terra Pro without maintaining a separate provider-specific connection, and keep the flexibility to test, switch, or combine models as your application requirements evolve.

01

One API integration

Access GPT-5.6 Terra Pro and other AI models through the same API workflow instead of maintaining separate integrations for every provider.

02

Easy model switching

Test GPT-5.6 Terra Pro against alternative models or switch models as your performance, capability, or cost requirements change without rebuilding your application around another provider API.

03

Flexible for production

Use GPT-5.6 Terra Pro from experimentation through production while keeping your AI stack flexible as workloads, traffic, and model requirements evolve.

04

Multi-model applications

Use GPT-5.6 Terra Pro for the workloads where it performs best and combine it with other models for tasks that require different capabilities, performance, or efficiency.

Frequently Asked Questions

Answers to common questions about integrating and using this AI model via AnyAPI.ai

GPT-5.6 Terra Pro is OpenAI's mid-tier GPT-5.6 Terra model served with reasoning.mode set to pro. It is the same underlying model as GPT-5.6 Terra but performs more internal reasoning per request for higher-quality answers on complex tasks, at the cost of higher latency and token usage. It was released July 9, 2026.

They are the same underlying model. The only difference is reasoning mode: standard Terra uses the default standard mode, while Terra Pro uses pro mode, which does more model work per request for better results on difficult tasks. Both share the 1,050,000-token context, 128,000-token output, modalities, and API features.

GPT-5.6 Terra Pro has a 1,050,000-token context window and supports up to 128,000 output tokens per completion. Prompts above 272K input tokens enter a higher long-context pricing tier, and a large window does not guarantee perfect retrieval of every fact across it.

GPT-5.6 Terra Pro accepts text, image, and PDF/document input and returns text only. It does not generate images, audio, or video. It supports function calling via tools and tool_choice, plus structured outputs through a JSON schema in response_format, and works best via the Responses API.

Use Terra Pro for genuinely hard tasks — complex coding, long-horizon agents, high-stakes extraction, or analytical review — where a more accurate answer justifies extra latency and token cost. For interactive UIs, high-throughput pipelines, or everyday work where standard quality suffices, standard GPT-5.6 Terra or the faster Luna tier are better.

* Benchmark data source: Artificial Analysis artificialanalysis.ai