Anthropic and OpenAI cut AI coding prices on the same day — what the Opus 5.5 vs GPT-6 Sol and Luna price war means for AI-assisted development costs
Anthropic's Claude Opus 5.5 and OpenAI's GPT-6 Sol and Luna both launched on 22 September 2026 with double-digit price cuts, the latest round in an AI coding tools price war that keeps lowering the cost of AI-assisted software development for the businesses commissioning it.
27 September 2026
On 22 September 2026, Anthropic released Claude Opus 5.5 and OpenAI answered on the same day with GPT-6 Sol and Luna — and both launches led with price, not just capability. Opus 5.5 is priced at $4/$20 per million input/output tokens, roughly 20% below Opus 5, with cache reads cut 60% to $0.20. GPT-6 Sol and Luna undercut that further: Luna at $0.10/$0.50 and Sol at $2/$10 per million tokens, both around 50% below GPT-5.6’s already-discounted rate, with a 90% discount on cached input reads. Neither company frames this as a temporary promotion — OpenAI confirmed the new rates carry no expiration date.
Capability didn’t take a back seat
This isn’t just a race to the bottom on price. Opus 5.5 leads the independent Artificial Analysis Intelligence Index at 58, scores 66.4% on Terminal-Bench 4.0, and became Claude Code’s default model the same day it launched, with five-hour session limits rising 20% — and going roughly 25% further per session at the new price. Sol and Luna carry over “the technological advancements behind GPT-6 Astra” into faster, cheaper models built for scaled, repeated use rather than occasional heavyweight tasks.
Why two companies cut prices on the same day
This is the clearest evidence yet that AI coding tools have entered a genuine price war rather than a slow, incremental decline. Anthropic and OpenAI are both betting that agentic coding — tools running longer, more autonomous sessions across a whole codebase — is where developer and business spend is heading, and neither wants to be the more expensive option for that workload. For a market that only recently settled into premium pricing for frontier models, two simultaneous cuts from the two largest labs signal that competition, not just efficiency gains, is now setting the price floor.
So what
If you’re budgeting an AI-assisted build, or you paused a project because the tooling cost didn’t pencil out six months ago, run the numbers again — the unit economics behind agentic AI coding have shifted twice in the last month alone (Fable 5.1’s cuts on 1 September, now this). That doesn’t change what actually matters in a build: engineering judgement, review discipline, and a team that treats these tools as a multiplier rather than a substitute for process. It does mean the budget conversation is worth revisiting. See how we scope AI-assisted development projects around tools like Opus 5.5, or get in touch to talk through what a build costs under today’s pricing.