Simon Willison: Claude and ChatGPT Hit Product-Market Fit

What exactly does Willison mean by "product-market fit" for Anthropic and OpenAI?

Simon Willison's May 27, 2026 post argued that Claude Code, Claude Cowork, and Codex have crossed the November 2025 inflection point. He measured PMF not by user counts but by enterprise willingness to lock into year-long contracts at higher prices. The thesis: token-burning workflows are now indispensable, so prices ratchet up without churn.

How much have API prices actually moved between GPT-5.4 and GPT-5.5?

GPT-5.5 prices at 2x the API price of GPT-5.4, per Willison's May 27 post. Claude Opus 4.7 prices at roughly 1.4x Opus 4.6 after tokenizer normalization. Both labs raised prices through new-model launches rather than direct hikes, which preserves the marketing line about old-model price stability. The economic effect is identical.

Why are enterprise customers willingly locking in at higher rates?

Switching cost. Coding agents and customer-facing assistants now sit inside core workflows. Replacing Claude Code with Codex requires retraining staff, rewriting prompts, and rebuilding orchestration. The cheapest way through a 2x price hike is to absorb it, since rebuilding from scratch costs more than the annual premium. Lock-in just got real.

When did the November 2025 inflection point happen and what changed?

Willison points to late November 2025 as the moment coding agents transitioned from experiment to production tool. Claude Code added background task execution. Codex got native git integration. Cursor hit Series C terms. The combined effect was suddenly daily usage rather than weekly experimentation. Three months later, the spend curve went vertical.

Where do coding agents fit into the new economics?

At the center. Willison wrote that coding agents "burn vastly more tokens" than chat usage, while serving extremely well-compensated professionals. The ratio is favorable to vendors. A senior engineer billing $200 per hour pays $50 per day for Claude Code without flinching. Multiply across thousands of engineers and the unit economics rewrite themselves.

Which Anthropic and OpenAI products are driving the revenue surge?

Anthropic: Claude Code, Claude Cowork, and Enterprise contracts. OpenAI: Codex, ChatGPT Pro, and ChatGPT Enterprise. OpenAI confirmed $2 billion in monthly revenue in its March 2026 $122 billion funding announcement. Anthropic's $1.25 billion monthly SpaceX bill is one downstream signal. Both companies are spending it about as fast as it arrives.

Should B2B SaaS founders renegotiate API contracts before July?

Yes if you have leverage, no if you do not. Founders renewing in July face the new pricing baseline. Lock annual rates in June where possible. If your usage runs steady at predictable volume, Anthropic Enterprise will commit a year. If your usage spikes, stay month-to-month and accept the premium. Avoid bridges between the two.

Will the SaaSpocalypse narrative hold or fade in H2 2026?

It will fade as PMF crystallizes. The SaaSpocalypse thesis assumed AI labs would commoditize vertical SaaS. Willison's post suggests the opposite: labs are extracting rent from specialized SaaS users rather than replacing them. B2B SaaS founders should compete on workflow depth, not feature breadth. The labs are too expensive to host your competitor.

Can a Bengaluru-based studio absorb a 2x API price jump without raising client rates?

Phoenix Studio's experience: yes, for one quarter. Beyond that, retainer rates need a small adjustment. The simplest path is a 12 to 18% retainer step in July, applied uniformly across all clients with a one-month notice. Most Bengaluru SaaS clients accept it when the rationale is named and the timing is transparent.

Does the IPO timing for both labs change founder negotiating leverage?

Yes. Anthropic and OpenAI both face IPO windows in late 2026 or early 2027. Pre-IPO labs need predictable revenue, which favors annual commitments. Founders who can offer multi-year volume commitments can extract custom pricing today that disappears after the S-1 prints. Q2 is the negotiating window for SaaS founders with leverage.

Modeling your H2 AI line item and need a frame for the price-up trajectory? OpenAI's C2PA push and the Mythos toggle sighting together hint at where production agents are headed. Let's chat.