Skip to content

OpenAI and Anthropic Slash Model API Prices in Same-Day Releases

Short answer

OpenAI released GPT-6 Sol and Luna at roughly half the price of their GPT-5.6 predecessors, while Anthropic cut Claude Opus 5.5 pricing 20% on input/output and 60% on cached tokens. For any company running AI agents or chatbots on these APIs, inference costs just dropped substantially without any migration effort beyond a model-name swap.

What this means for operators

If your support bot, sales assistant or internal automation runs on GPT or Claude APIs, this is a direct cost lever: GPT-6 Luna at $0.10/M input tokens is roughly one-twentieth the price of Claude Opus 5.5, and cached-token discounts on Opus 5.5 matter a lot if your agents run long multi-turn conversations where most tokens are repeated context. Before renewing any AI vendor contract or locking in a model choice for a new automation build, re-run the cost math — a workflow that looked expensive at GPT-5.6 or Opus 5.0 pricing may now be cheap enough to expand to more use cases, and a support queue currently routed to a premium model may run just as well on a cheaper one.

OpenAI and Anthropic released new models within an hour of each other on 22 September 2026, triggering what Simon Willison calls a price war among frontier AI vendors.

OpenAI's GPT-6 Sol and GPT-6 Luna arrived priced at roughly half their GPT-5.6 equivalents. GPT-6 Luna now costs $0.10 per million input tokens and $0.50 per million output tokens — among the cheapest models OpenAI has shipped, beaten only by smaller Nano-tier models. GPT-6 Sol dropped to $2/$10 per million tokens, matching what GPT-5.6 Terra used to cost, effectively removing any reason to keep using Terra. Willison also notes GPT-5.6 pricing was scheduled to rise 25% in November, meaning GPT-6's discount looks even larger against that baseline.

Anthropic's Claude Opus 5.5 got a 20% price cut versus Opus 5.0 through 4.8, moving from $5/$25 to $4/$20 per million input/output tokens. Cache-read pricing fell 60%, which matters directly for agentic workloads where, per Willison, over 90% of input tokens in longer conversations are served from cache. Anthropic also said cheaper Sonnet 5.5 and Haiku 5.5 models are coming, and specifically flagged Haiku's need to compete with GPT-6 Luna's pricing.

Grok 4.7, released the prior day, now sits between the two: it undercut GPT-5.6 Sol substantially at launch but is now roughly matched by GPT-6 Sol on input pricing.

Willison reports mixed results testing the models with his standard pelican-drawing benchmark: Claude Opus 5.5 at its "max" reasoning setting failed twice, hitting the 128,000-token output limit while still reasoning about a simple SVG request, costing $2.56 and about 20 minutes per failed attempt. He has since switched his own default coding tools to GPT-6 Sol and Claude Opus 5.5, and upgraded a public Datasette demo to GPT-6 Luna.

Source: Simon Willison

Next step

Visibility Analyzer

This is what the Visibility Analyzer measures on a real site: which answers cite you, which pages an engine cannot retrieve, and what to fix first. Free to run.

Run a free visibility audit

Free to run. No card.

Fee
Free
Length
One run, minutes

Free tier: two analyses a day, no card required.