AI & Tech

Claude Opus 5.5 vs. GPT-6 Sol/Luna: Same-Day Price Cuts, Neither Official Chart Lists the Other's New Model

On September 22, 2026, Anthropic and OpenAI both cut prices on the same day. Neither company's official comparison chart lists the other's new model from that day, so the charts cannot settle which new release wins. This piece lays out what each company's pricing and comparison charts actually show, where they stop, and what's actually worth comparing as a reader.

2026.09.23 · By dvdmaru · ~5 min read · 1,165 words

中文版

Tuesday, September 22, 2026. Anthropic released Claude Opus 5.5, and OpenAI released GPT-6 Sol and GPT-6 Luna. Same day, two companies, both moving their price lists down a notch.

Which of the two new models is stronger? The official comparison charts can’t tell you. Lay each company’s own chart next to the other and it shows up directly: Anthropic’s chart has no GPT-6 Sol, no GPT-6 Luna. OpenAI’s chart has no Claude Opus 5.5. Two documents published the same day, and neither lists the other’s new model.

That is what the charts show. It does not show that neither company could have tested against the other’s new model beforehand, and an absence in a chart is not evidence of why. What can be said is narrower: neither official chart can decide the winner between the two same-day releases.

What actually got cheaper

Opus 5.5’s pricing is straightforward: $4 and $20 per million tokens for input and output, 20% below Opus 5. Cache reads drop to $0.20 per million, 60% below Opus 5. Anthropic says that at default settings, typical-workload costs fall 40% overall, and output generation runs more than 30% faster. That 40% is Anthropic’s own estimate for an overall workload, separate from the per-token cuts and measured on a different basis.

Anthropic frames Opus 5.5 as “the first model in our new Claude 5.5 family”, matching the tier above it, Fable 5.1, on most work while costing 40% less than Opus 5 to run.

OpenAI positions GPT-6 Sol and Luna below its flagship, GPT-6 Astra, as the value tier — Astra launched September 3 as a new model, with no price cut attached to that release. Sol’s pricing goes from $4 to $2 on input and $20 to $10 on output, an exact 50% across the board. Luna’s input also halves, from $0.20 to $0.10, but its output drops from $1.20 to $0.50 — a 58% cut, more than the press release’s blanket “50% cheaper” line. That reduction is measured against GPT-5.6’s promotional pricing, not list pricing.

Claude Opus 5.5GPT-6 SolGPT-6 Luna
Released2026-09-222026-09-222026-09-22
Input (per million tokens)$4 (-20%)$2 (-50%)$0.10 (-50%)
Output (per million tokens)$20 (-20%)$10 (-50%)$0.50 (-58%)
Cache reads$0.20 (-60%)Only a 90% discount disclosed, no absolute priceSame
Speed / positioning-40% typical cost, +30% output speed~Half the error rate of predecessor on internal factuality evalMatches GPT-5.6 Sol at ~1/100th its cost at higher effort on that internal factuality evaluation

The cuts are measured against different baselines: Opus 5.5 against Opus 5, Sol and Luna against GPT-5.6’s promotional pricing. The figures in the bottom row are each company’s own claims, not third-party measurements.

Who each company benchmarks against

The pricing table above is uncontroversial — both companies published those numbers directly. The performance comparisons are where the gap is: neither company’s chart includes the other’s new model from that day.

Anthropic’s comparison chart lists five models: Opus 5.5, Fable 5.1, Opus 5, GPT-6 Astra, and GPT-5.6 Sol — not one cell mentions GPT-6 Sol or GPT-6 Luna. The clearest illustration is on CursorBench: Anthropic writes that Opus 5.5 “beats GPT-5.6 Sol by 11 points for about a third of the cost”, but that “11 points” figure uses default (medium) effort scores. The main table’s max/xhigh scores put the gap at 57.8% versus 41.7% — a 16.1-point difference. Same benchmark, different effort level, different number. The Terminal-Bench 4.0 methodology note says the GPT-6 Astra and GPT-5.6 Sol figures for that benchmark are as reported by OpenAI, not independently reproduced.

OpenAI’s chart runs the same way in reverse: not one cell mentions Claude Opus 5.5. On AutomationBench, GPT-6 Sol at xhigh effort scores 33.2% at $0.27 per task, against “Claude Fable 5.1 w/ Opus 5 Fallback” at 31.4% — more than 8.9x Sol’s cost, with the fallback’s own cost left out of that number; OpenAI’s own text notes the fallback triggered on roughly 40% of tasks. OpenAI’s methodology note says the same kind of thing as Anthropic’s: competitor scores were taken from publicly available reports, not run in-house.

Both charts do something similar: each company measures its newest model against models the rival had already published. Why the same-day models are missing, neither document says, and this piece does not guess. The result is that neither of these two “same-day” documents constitutes a real head-to-head test of what actually launched that day.

What’s actually worth comparing

Read only the headlines and this “price war” sounds like two companies going head-to-head, racing each other on speed and cost. Look at what the documents actually contain, though: each company published its prices and its own comparison chart, and neither chart includes the other’s same-day model.

A cut in list price also doesn’t translate into a proportional cut in what a real task costs. List price, cache hit rate, token consumption, retries and latency are different things, and the same job can burn a different number of tokens on a different model.

As for which company to actually use, neither official comparison chart can answer that. What can answer it is the workload in front of you — how many tokens it burns, whether it can tolerate retries, how much of the bill is cache hits. The price lists will keep moving. Run your own task once, and don’t take a chart posted on launch day as the conclusion.

Frequently asked questions

How much cheaper are GPT-6 Sol and Luna than GPT-5.6?

OpenAI’s press release says 50% cheaper than GPT-5.6’s promotional pricing. Check the table column by column and Sol’s input and output are exactly 50% off. Luna’s output, though, drops from $1.20 to $0.50 per million tokens — that’s actually a 58% cut, more than the headline number.

So which model is actually stronger, Claude Opus 5.5 or GPT-6 Sol/Luna?

Neither company’s own launch-day comparison chart can answer that, because neither lists the other’s new model from that day. Anthropic’s chart benchmarks against GPT-6 Astra and GPT-5.6 Sol — no GPT-6 Sol or Luna. OpenAI’s chart benchmarks against Claude Opus 5 and Fable 5.1 — no Claude Opus 5.5. Neither document explains why. Neither is a real head-to-head test of what actually launched that day.