Infrastructure · AUGUST 30, 2026
The Frontier API Price Floor Just Dropped Under Everyone's Q4 Budget
Anthropic made Sonnet 5's $2/$10 introductory rate permanent, cancelling a September 1 hike to $3/$15. Combined with OpenAI's GPT-5.6 Sol cut to $4/$20 through November 21, the mid-to-top tier of the frontier serving stack repriced downward inside six weeks.
Anthropic quietly cancelled the September 1 price hike that would've made Claude Sonnet 5 roughly 50% more expensive, appending an edit to its June 30 launch post on August 10 that made the $2/$10 per-million-tokens introductory rate permanent. Two weeks later, OpenAI cut GPT-5.6 Sol from $5/$30 to $4/$20 and guaranteed that rate through at least November 21, 2026. Inside six weeks, the mid-to-top tier of the frontier serving stack repriced downward.
The scoreboard shift matters more than either move on its own. GPT-5.6 Sol at $4 input and $20 output now undercuts Claude Opus 5, which holds at $5/$25 alongside Opus 4.8. For the first time, OpenAI's most capable general-purpose tier is cheaper than Anthropic's on both sides of the ledger. GPT-5.6 Terra sits below at $2/$12 short-context, and Sonnet 5's frozen $2/$10 anchors the working-horse tier that most production applications actually run on.
For small-team operators, the practical read is simple: the Q4 budget line you were bracing for isn't coming. Vendors building customer-acquisition, outreach, and content tools on Sonnet 5, Sol, or Terra saw their cost of goods either drop or hold flat. That's the input side of a lot of AI SaaS. Whether those savings pass through to seat prices is a renewal conversation, not an assumption.
One caveat worth writing into the same conversation. Anthropic's platform docs, flagged in Enterprise DNA's write-up, note that Sonnet 5's newer tokenizer produces roughly 30% more tokens for the same text than Sonnet 4.6 and earlier. Sonnet 4.6 lists at $3/$15. So the effective per-request delta between the two is narrower than the sticker gap suggests. It's still cheaper. It's not two-thirds cheaper.
Two structural notes sit under all of this. The first is that OpenAI's Sol cut landed weeks after Claude Opus 5 shipped, and Anthropic's permanence decision followed. The lab that moves last on price sets the floor, and both labs are now moving reactively rather than defending a premium. The second is that permanence is doing real work here. A promotional rate is a Q4 wildcard. A permanent rate is a line in the budget.
The September budget adjustment small operators penciled in around Sonnet 5 doesn't need penciling anymore. That's the whole story, and it's the kind of story that only reads as small until renewal week.