GPT-6 Luna vs GPT-6.1 Sol: 14 Times the Price, What Does Sol Buy?

Two OpenAI models, one board deck, the same data and the same pipeline. The bigger model costs 14 times as much per deck. Here is what the difference actually got us.

By Ed · Published November 5, 2026 · 4 min read

Twenty-five years starting companies, consulting, owning a multi-million-dollar P&L and leading business intelligence for large enterprises.

Sol costs 14 times as much per deck ($2.08 against 15 cents) and is faster (5.4 minutes against 8.6). What it buys is analysis: campaign efficiency and loss reasons over time, where Luna summarized where the money went. Sol found 3 of 8 story points in our checklist, Luna 1. Neither reached the Southeast win-rate story.

How this was tested. One prompt, “Northvale 2026 year-to-date board review, with an executive summary and recommendations”, run once per model in the Quiriz PowerPoint add-in on November 5, 2026. Northvale is a fictional US apparel company with stories planted in its data: Google and Meta ad exports and a HubSpot wholesale pipeline, through September 15, 2026. Every step ran on the model under test: plan, build (answering each slide’s question from the data), gap analysis, write and critic; see how a deck is built. Three small parts stayed fixed for every run (chart styling, question routing and a grounding check). Cost is what OpenRouter billed. Build time runs from pressing Build to the last slide written, and leaves out the plan, which waits for a person to approve it. One run per model, so treat close results as a tie. Full method on the benchmark page.

At a glance

	GPT-6 Luna	GPT-6.1 Sol
Billed cost	$0.15	$2.08
Build time	8.6 min	5.4 min
Analysis calls	47	33
Content slides	11	11
Stories found	1 of 8	3 of 8
Meta prospecting	Where the money went	Cost up 50%, conversions flat
Southeast	The decline, not the reason	Not covered
Recommendations	Back the growing regions	Pricing, buyer commitments, a NonBrand test

Where they agree

Both decks lead with the same shifts: Prospecting-Lookalike took 82.3% of Meta’s added spend and PMax-Retarget 61.2% of Google’s, and wholesale won amount rose $299K.

What the extra $1.93 bought

Efficiency, not just spend. Sol set conversions against cost for every campaign: Lookalike cost up 50% with conversions flat, NonBrand-Search conversions up 21.4% on 12.1% more cost. Luna stopped at where the money went.
Loss reasons over time. Sol showed pricing losses up $104.7K and deals going dark up $150.8K year over year.
Recommendations a board could use. Luna recommended more effort on the two regions already growing; Sol ranked three specific actions with owners.

GPT-6.1 Sol deck: ranked priorities: pricing, buyer commitments, then budget reallocation — GPT-6.1 Sol’s recommendations: three ranked actions with owners.

What neither found

The Southeast won 4 of its 20 closed deals in 2026, against more than half everywhere else, and lost mostly on price. Luna reported the Southeast decline without asking why. Sol looked at losses by region, saw them roughly even, and did not set them against wins. Claude Opus 5.5 found it in one slide.

Verdict

If the deck is a summary you will annotate yourself, Luna at 15 cents is fine. If the deck is the analysis, Sol is clearly better and faster. But on this test the better buy below $1 is neither: DeepSeek V4.1 Flash found more than Sol for 30 cents.

Run the same deck on your own data

Describe the deck in PowerPoint and see the source of every figure. Free to start.

Try Quiriz free →

Frequently asked questions

Is GPT-6.1 Sol better than GPT-6 Luna?

For analysis, yes. On the same deck Sol set ad conversions against cost for every campaign and ranked loss reasons over time, finding 3 of 8 story points in our checklist against Luna’s 1. For a summary of totals, Luna is enough.

How much more does GPT-6.1 Sol cost than Luna?

About 14 times per deck in our test: $2.08 against 15 cents. List prices are $2 and $0.10 per million input tokens, a factor of 20; caching narrowed the gap.

Which is faster, Luna or Sol?

Sol: 5.4 minutes against 8.6, because it made 33 analysis calls to Luna’s 47.

Is there a better cheap option than Luna?

On this test, DeepSeek V4.1 Flash at 30 cents found the ad problem neither OpenAI model followed through on, and 5 of 8 story points.