The GPT-5.6 Family Is Now in Kolbo: Sol, Terra, and Luna

OpenAI shipped the GPT-5.6 family this week. It is not a single model. It is a three-tier system designed to cover the full range of use cases from high-volume everyday tasks to frontier-level professional work, all sharing the same core architecture: 1.05 million token context, 128K max output, text and image input, and a knowledge cutoff of February 2026.

All three are live in Kolbo Chat right now.

The Three Tiers at a Glance

GPT-5.6 Sol: The Flagship

Sol is OpenAI's frontier-tier GPT-5.6 model. It is the one you reach for when the task has no room for approximation: complex research synthesis, multi-document analysis, detailed architecture reviews, difficult debugging sessions, or any work where you need to feel confident that the model has genuinely thought through the problem.

What makes Sol different from the previous generation: reasoning is tunable per request. Set the effort to low for fast, direct answers. Set it to high and Sol applies deep, multi-step internal reasoning before committing to a response. You stay in one model, in one conversation, and adjust as the task demands.

At $5 / $30 per 1M tokens, Sol is priced at the same level as the previous GPT-5.5, which means you are getting materially better reasoning capability for no additional cost.

Go directly to Sol: app.kolbo.ai/chat?model=gpt-5.6-sol

GPT-5.6 Terra: The Practical All-Rounder

Terra sits in the middle of the family. OpenAI describes it as the balance between intelligence and cost. In practice, it runs the same 1.05M context and 128K output as Sol, with the same image input support, at half the price.

For most everyday professional work, Terra is the right default. Writing, editing, summarizing long documents, explaining code, answering complex questions, building detailed outlines. The tasks where you want GPT-5.6 quality but do not need the deepest reasoning Sol can produce.

At $2.50 / $15 per 1M, Terra is the practical choice for regular use without burning through credits on tasks that do not require frontier-tier compute.

Go directly to Terra: app.kolbo.ai/chat?model=gpt-5.6-terra

GPT-5.6 Luna: Cost-Optimized at Scale

Luna is the high-volume tier. At $1 / $6 per 1M tokens, it is by far the most economical way to access GPT-5.6 capabilities. It carries the same 1.05M context, 128K max output, and image input as its siblings. The difference is in compute: Luna is optimized for throughput over depth.

Where Luna fits: any workflow that calls the model frequently, tasks where volume matters more than maximum reasoning depth, draft generation before human review, quick summaries, classification, and extraction.

Go directly to Luna: app.kolbo.ai/chat?model=gpt-5.6-luna

The One Thing All Three Share

Every model in the GPT-5.6 family includes reasoning tokens. Unlike previous generations where reasoning was a separate mode or a separate model, reasoning is part of the GPT-5.6 architecture across all three tiers. The difference between Sol, Terra, and Luna is not whether they reason. It is how deeply, and at what cost.

Combined with 1.05 million token context, all three can hold an entire codebase, a year of documents, or a lengthy research corpus and reason across the full input without losing context from early parts.

What to Use Each Model For

For deep analysis and high-stakes work: Sol

Architecture reviews, legal document analysis, complex debugging, research synthesis, tasks where the cost of a shallow answer is a problem downstream. High reasoning effort at Sol's depth is worth the price when the output matters.

For everyday professional tasks: Terra

Writing, editing, code explanation, document summarization, Q and A with complex context. The tasks you do daily that benefit from GPT-5.6 quality without needing Sol's full compute every time.

For high-volume and cost-sensitive work: Luna

Drafts, summaries, classification, extraction, any workflow that calls the model frequently. Luna gives you the GPT-5.6 architecture at a price that makes scale practical.

For mixed workflows: pick per message

In Kolbo Chat you can switch models mid-conversation. Start a research session in Sol, move to Terra for drafting, and use Luna for quick checks. Your conversation history carries over.

Frequently Asked Questions

What is the GPT-5.6 family?

GPT-5.6 is a three-model family from OpenAI released in July 2026. All three models share the same core: 1.05 million token context, 128K max output, text and image input, reasoning tokens, and a knowledge cutoff of February 2026. The tiers are Sol (frontier, $5/$30 per 1M), Terra (mid-tier, $2.50/$15 per 1M), and Luna (cost-optimized, $1/$6 per 1M).

What is the difference between GPT-5.6 Sol, Terra, and Luna?

The three tiers represent a tradeoff between reasoning depth and cost. Sol is the most capable and designed for complex professional work. Terra balances intelligence and cost for everyday professional tasks. Luna is optimized for high-volume workloads where throughput matters more than maximum depth. All three have the same context window, output cap, and image input support.

How do I use GPT-5.6 online without a subscription?

You can access all three GPT-5.6 models through Kolbo.AI, which has a free tier. Go to app.kolbo.ai, create a free account, and select the model from the Chat model picker. Use code KOLBO30 for 30% off your first paid month.

What is the difference between GPT-5.6 Sol and GPT-5.5?

GPT-5.6 Sol adds tunable reasoning effort tiers, a 1.05 million token context window (significantly larger than 5.5), 128K max output, and native image input. The price is comparable to GPT-5.5, making Sol a strict upgrade for most tasks.

What does "reasoning tokens" mean in GPT-5.6?

Reasoning tokens are internal compute steps the model takes before producing its visible answer. The model thinks through assumptions, edge cases, and intermediate steps, then responds. In Sol, you can influence how many of these steps happen by setting reasoning effort (low, medium, or high). In Terra and Luna, the effort is balanced or optimized automatically.

Which GPT-5.6 tier should I use?

If you are doing work where accuracy and depth matter more than speed or cost, use Sol. If you are doing everyday professional tasks and want the best balance of quality and credit efficiency, use Terra. If you are running high-volume workflows or doing tasks that do not require maximum reasoning depth, use Luna. When in doubt, start with Terra.

What is the best OpenAI model in 2026?

As of July 2026, GPT-5.6 Sol is OpenAI's most capable publicly available model for reasoning and multimodal tasks. For most professional users who do not always need frontier-level compute, GPT-5.6 Terra is the practical recommendation.

Can the GPT-5.6 models analyze images?

Yes. All three GPT-5.6 models support native image input. You can attach screenshots, diagrams, charts, and photos directly to your Chat message in Kolbo, and the model will read and reason over them in the same response.

Is the GPT-5.6 family available in Kolbo?

Yes. All three models are fully integrated into Kolbo Chat. Open the Chat tool, select the model you want, and start. Direct links: Sol, Terra, Luna.

Best, Zohar Founder, Kolbo.AI