OpenAI GPT-5.6 Sol: One tweet. One price comparison. One very loud message to Anthropic.
Sam Altman doesn’t do subtle. The OpenAI CEO took to X this week with a pointed message: GPT-5.6 Sol is half the price and roughly twice as token-efficient as Fable in many cases for accomplishing the same task — and at medium reasoning, it performs comparably at one-quarter of the estimated cost. That’s not just a product claim. That’s a direct challenge to Anthropic’s flagship model, issued in public, by name.
Here’s the full picture behind the numbers.
Table of Contents
What Is GPT-5.6, and What Are Sol, Terra, and Luna?
GPT-5.6 comes in three variants: Sol (considered its workhorse), Terra (a more intermediate option), and Luna (its budget-friendly option). The full family launched publicly on July 9, 2026, after a government-coordinated limited preview that began June 26.
GPT-5.6 is set to be the operating agent of ChatGPT Work, a service released at the same time. The two launches are deliberately paired — GPT-5.6 is the engine; ChatGPT Work is the product built on top of it. We covered ChatGPT Work separately — it’s worth reading alongside this piece for the full picture of what OpenAI shipped this week. You can find that and more in our AI tools and technology coverage at TechnoSports.
GPT-5.6 Sol vs Claude Fable 5: The Numbers
This is the comparison everyone is running right now. Here’s how they stack up:
| Metric | GPT-5.6 Sol | Claude Fable 5 |
|---|---|---|
| Input price (per 1M tokens) | $5 | $10 |
| Output price (per 1M tokens) | $30 | $50 |
| Cost per task (AA Index, max) | $1.04 | $2.75 |
| Coding Agent Index score | 80 (top) | 77 |
| Token efficiency (agentic coding) | 54% more efficient | Baseline |
| Task time | 61% less | Baseline |
Cost per task in the Artificial Analysis Intelligence Index: GPT-5.6 Sol (max) is $1.04 — about one-third of the cost of Claude Fable 5 (max) at $2.75. The more affordable variants, Terra and Luna, cost $0.55 and $0.21 per task, respectively.
The real-world gap is even starker. One developer built a café ordering app on each model and paid $7 with Sol versus $62 with Fable. That’s not a marginal difference — it’s a nine-times cost gap on an identical task.
What Sol Is Actually Good At
OpenAI calls Sol its “best coding model yet,” claiming that Sol sets a new state of the art at 80 on the Coding Agent Index, 2.8 points above Fable 5, while using less than half the output tokens, taking less than half the time, and costing about one-third less.
Sol’s benchmark performance on Terminal-Bench 2.1 — 88.8% in standard mode, 91.9% in ultra — comes from a specific architectural feature: ultra mode is not simply “more compute applied to the same reasoning chain.” It is a multi-agent system embedded in the model itself. When a user runs a request in ultra mode, Sol decomposes the task and spawns parallel subagent processes, each working on a different component simultaneously.
The efficiency claim extends beyond code. OpenAI calls 5.6 its “strongest cybersecurity model yet, achieving frontier performance with significantly fewer tokens.”
Where Fable 5 Still Holds Ground
The pricing story is clear. The performance story is more nuanced.
According to the independent AI evaluation platform Artificial Analysis, GPT-5.6 Sol (max) scores 59 points on the Intelligence Index, just one point behind Claude Fable 5 (max). That’s near-parity on general intelligence — but Fable 5 still leads overall.
For complex coding and finished product quality, some testers still felt Fable 5 was slightly ahead. In creative coding and one-shot prototyping, the final user experience can matter more than the model’s raw benchmark number.
The honest summary: Sol wins on price and coding benchmarks; Fable 5 retains an edge in creative and complex one-shot tasks. For enterprises running high-volume agentic workflows, Sol’s cost advantage is likely decisive. For developers who care most about output quality on complex creative problems, it’s less clear-cut.
The Bigger Picture: AI Is Getting a Lot Cheaper
Skyrocketing token costs have become the newest challenge to AI adoption. Uber made headlines earlier this year for blowing through its entire AI coding budget for 2026 in just four months, while Microsoft started cancelling its developers’ Claude Code licences months after enabling them.
GPT-5.6’s pricing directly addresses that problem. Pricing pressure is building across the AI industry. Chinese open models, Meta with Muse 1.1, and xAI with Grok 4.5 are all pushing established AI labs to cut rates. OpenAI is still pricier than those competitors but is now squeezing Anthropic from another angle.
For Indian developers and enterprises evaluating AI tools, the cost-per-task gap between Sol and Fable 5 is now substantial enough to factor into architecture decisions. Our best AI tools for developers and enterprises in India roundup covers the full landscape.
Bottom Line
Sam Altman’s tweet wasn’t boasting — it was a pricing memo. GPT-5.6 Sol is 54% more token-efficient on agentic coding tasks and “as good or better” than competing models on the market at roughly half the price. For cost-conscious enterprises, that’s a compelling case to switch or at least run parallel evaluations. Whether Sol can match Fable 5’s quality ceiling on the hardest tasks is still an open debate — but at these prices, it doesn’t need to win every comparison to win the market.
Sources: CNBC | TechCrunch | The Decoder
