GPT-5.6 Explained: What's New in Sol, Terra & Luna

GPT-5.6 is OpenAI's biggest naming shake-up in years, and it's not one model — it's three. If you've been confused by references to "Sol," "Terra," and "Luna" showing up in ChatGPT, here's exactly what GPT-5.6 changes, what each tier costs, and which one is actually worth using.

Unlike previous releases where OpenAI shipped a single model with a reasoning dial, GPT-5.6 arrives as three distinct, independently-updatable tiers. That's a real shift in how OpenAI plans to ship models going forward.

Table of Contents

• What Is GPT-5.6?
• Why This Release Matters
• The Three Tiers: Sol, Terra, Luna
• Release Timeline & the Government Delay
• Pricing (Updated July 30, 2026)
• Benchmarks: How It Compares
• GPT-5.6 and ChatGPT Work
• Which Tier Should You Use?
• TechZila Analysis
• Frequently Asked Questions

What Is GPT-5.6?

GPT-5.6 is OpenAI's newest large language model family, released generally on July 9, 2026. Instead of shipping one flagship model, OpenAI split the release into three durable tiers — Sol, Terra, and Luna — each aimed at a different point on the cost-to-capability curve. It replaces GPT-5.5 as OpenAI's frontier model line.

Featured Snippet Answer — What's new in GPT-5.6?

GPT-5.6 replaces a single flagship model with three separate tiers: Sol (most capable, tuned for coding, science, and cybersecurity), Terra (balanced, roughly Sol's little brother for everyday work), and Luna (fastest and cheapest). OpenAI also introduced a new "Ultra" multi-agent mode, more predictable prompt caching, and a companion product called ChatGPT Work.

Why This Release Matters

Every past GPT release was "one model, choose your reasoning effort." GPT-5.6 changes that logic — Sol, Terra, and Luna are now meant to be ongoing product lines that can each improve on their own schedule, rather than being replaced wholesale every time OpenAI ships a new generation number. If that holds, it changes how developers plan long-term AI infrastructure around OpenAI's models.

The Three Tiers: Sol, Terra, Luna

• Sol (flagship): OpenAI's most capable model, built for long-horizon coding, scientific research, and cybersecurity work. OpenAI calls it its "best coding model yet" and its "strongest cybersecurity model yet," supporting defensive work like threat modeling, code review, patching, and blue teaming.
• Terra (balanced): Positioned as roughly comparable to GPT-5.5's performance at about half the cost — the default migration target for teams currently running GPT-5.5 workloads.
• Luna (fast & affordable): The cheapest, fastest tier, aimed at high-volume tasks like classification, tagging, and extraction where per-request cost matters more than peak capability.

OpenAI also introduced an Ultra mode within Sol — a multi-agent setting that coordinates several agents in parallel to work through harder, longer tasks faster, currently available to ChatGPT Work Pro/Enterprise users and Codex Plus-and-higher plans.

Release Timeline & the Government Delay

GPT-5.6's rollout wasn't straightforward. OpenAI first previewed the models on June 26, 2026, but limited access to a small group of trusted partners at the request of the U.S. government, which wanted to review the models' safety — specifically around their cybersecurity and biology capabilities — before wider release. OpenAI has said it does not believe this kind of pre-release government review should become the long-term default. The models became generally available across ChatGPT, Codex, and the API on July 9, 2026.

Pricing (Updated July 30, 2026)

Tier	Input (per 1M tokens)	Output (per 1M tokens)	Best For
Sol	$5.00	$30.00	Long-horizon coding, science, cybersecurity
Terra	~$2.00 (after 20% cut)	~$12.00 (after 20% cut)	Everyday production work, GPT-5.5 replacement
Luna	~$0.20 (after 80% cut)	~$1.20 (after 80% cut)	High-volume classification, tagging, extraction

On July 30, 2026, OpenAI cut Luna's price by 80% and Terra's by 20%, while holding Sol flat. The release also introduced more predictable prompt caching — cache writes cost 1.25x the uncached input rate, cached reads get a 90% discount, and cache life has a 30-minute minimum with support for explicit cache breakpoints. A new Fast mode in the API replaces the old Priority Processing option, delivering up to 2.5x faster speeds for Sol at double the price.

Benchmarks: How It Compares

OpenAI's own charts show Sol setting new highs on BrowseComp (92.2%) and OSWorld 2.0 (62.6%), beating Opus 4.8 on OSWorld while using 85% fewer output tokens. On Terminal-Bench 2.1, an agentic coding benchmark, Sol scored 88.8% (91.9% in Ultra mode), narrowly ahead of Claude Mythos 5 and GPT-5.5, both at 88.0%.

The picture isn't one-sided, though. Independent evaluations cited by third-party trackers show Claude Mythos 5 and Claude Fable 5 leading on other benchmarks, including Toolathlon and GraphWalks, and GPT-5.5 narrowly edging out Sol on one long-context MRCR test. Luna, notably, beats Claude Opus 4.8 on two coding benchmarks (74.6 vs. 72.5 on the Artificial Analysis Coding Agent Index, and 67.2% vs. 59.0% on DeepSWE) at a fraction of the cost.

GPT-5.6 and ChatGPT Work

GPT-5.6 launched alongside ChatGPT Work, a new enterprise-focused product that runs on Sol as its operating agent. It's designed to take messy context from documents, Slack, Notion, Microsoft 365, and Google Drive, and convert it into polished, shareable artifacts — turning natural-language requests into interactive explanations and visualizations.

Which Tier Should You Use?

Your Need	Recommended Tier
Long-horizon coding across files, tests, and fixes	Sol
Migrating existing GPT-5.5 workloads on a budget	Terra
High-volume classification, tagging, or extraction	Luna
Cybersecurity or scientific research tasks	Sol
Everyday assistant use in ChatGPT	Terra

⭐ TechZila Verdict: 8.7/10

The tiered approach is a genuinely useful shift for developers who were previously stuck picking one reasoning effort. Luna in particular stands out as an aggressive price-to-performance play. Sol's edge over rivals is real but narrower than OpenAI's own marketing suggests once you look at independent benchmarks.

Best for: developers running production AI workloads, coding-heavy teams, and businesses evaluating ChatGPT Work for document-heavy workflows.

People Also Ask

What happened to GPT-5.5?
GPT-5.6 replaces GPT-5.5 as OpenAI's frontier model line; Terra is positioned as a like-for-like replacement at roughly half the cost.

What is Ultra mode?
Ultra is a multi-agent setting within Sol that coordinates multiple agents in parallel to complete complex, long-horizon tasks faster, available in ChatGPT Work and Codex on higher-tier plans.

Is GPT-5.6 available to free ChatGPT users?
Availability varies by plan and has changed since launch — check OpenAI's current ChatGPT pricing page for which tiers are included at each subscription level.

TechZila Analysis

The most interesting part of GPT-5.6 isn't a benchmark number — it's the tier system itself. OpenAI is betting that developers want durable product lines they can build around, rather than a single model that gets fully replaced every few months. Whether that promise holds depends on OpenAI actually updating Sol, Terra, and Luna independently going forward, instead of just shipping GPT-5.7 and starting the naming cycle over. The July 30 price cuts on Terra and Luna, without touching Sol, are an early signal that OpenAI intends to keep evolving the cheaper tiers on their own cadence — which is exactly what the tier model promised.

Related Reading

• 10 Best Free AI Tools in 2026
• Kimi K3 AI Review 2026
• Claude AI Chats Google Search Exposure Explained
• More AI News on TechZila

Frequently Asked Questions

What is GPT-5.6?
GPT-5.6 is OpenAI's newest model family, released as three separate tiers — Sol (flagship), Terra (balanced), and Luna (fastest and cheapest) — instead of one model with adjustable settings, replacing GPT-5.5 as OpenAI's frontier release.

How much does GPT-5.6 cost?
As of OpenAI's July 30, 2026 price update, Sol costs $5 input / $30 output per million tokens, Terra costs about $2 input / $12 output after a 20% cut, and Luna costs about $0.20 input / $1.20 output after an 80% cut. Prices can change, so check OpenAI's official pricing page.

Why was GPT-5.6 delayed at launch?
OpenAI first released GPT-5.6 as a limited preview to a small group of trusted partners on June 26, 2026, at the request of the U.S. government, which wanted to assess the safety of the model's cybersecurity and biology capabilities before wider release. It became generally available on July 9, 2026.

Is GPT-5.6 better than Claude?
It depends on the benchmark. OpenAI's own charts show Sol and Luna beating some Claude models on select coding benchmarks, while independent evaluations show Claude Mythos 5 and Claude Fable 5 leading on other benchmarks like Toolathlon and GraphWalks. There's no single universal winner across every task.

Official Sources

• OpenAI — GPT-5.6 Official Announcement
• OpenAI — Previewing GPT-5.6 Sol
• OpenAI — Official Release Notes & Pricing Updates

If your team is choosing between OpenAI's new tiers or comparing them against Claude and Gemini for a production workload, understanding these cost and capability trade-offs now will save a re-architecture later. Follow TechZila for the latest AI model releases, comparisons, and practical technology guides.