Anthropic and OpenAI both released new flagship-tier models on Tuesday — within minutes of each other — escalating a price war that has become the defining business story of the AI industry's second half of 2026. Anthropic's Claude Opus 5.5 delivers performance comparable to its top model at roughly 40% lower cost, while OpenAI counterpunched with GPT-6 Sol and GPT-6 Luna, two significantly cheaper members of its GPT-6 family now generally available on Amazon Bedrock.
The same-day launches, reported by CNBC, Fortune, Engadget and Decrypt, are the first major releases from either lab since a reported slowdown agreement between the companies became the subject of an antitrust lawsuit earlier this month — a fact that adds a competitive edge to what would already have been a consequential shipping day. For deeper numbers and analysis, see our AI industry coverage.
Claude Opus 5.5: Fable-Level Performance at 40% Less
In its announcement, Anthropic positioned Opus 5.5 as the first model of a new Claude 5.5 family that "performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5." The company says the model was tested before release by external evaluators including Frontier Design and METR, and that it posted the strongest results to date on Anthropic's automated behavioral audit, the most comprehensive alignment test the lab runs.
The pricing details matter as much as the benchmarks. Opus 5.5 costs $4 per million input tokens and $20 per million output tokens — 20% below Opus 5 — while cache reads, which Anthropic says make up the majority of agentic and coding workload costs, drop 60% to $0.20 per million tokens. Output generation is more than 30% faster than Opus 5, and a fast mode in Claude Code and the Claude Platform reaches 2.5x speed at $8 and $40 per million tokens respectively.
Anthropic's customer examples illustrate why efficiency is the headline. One early tester completed a 680,000-line code migration in less than a day — work the company says would have taken an engineering team weeks. Another audited and fixed a 200,000-line codebase in under three hours, where Opus 5 took over 20 hours and 2.5 times as many tokens. In an internal test translating the HAProxy load balancer from C to Rust, Opus 5.5 finished in 9.5 hours versus 12 for Fable 5.1 at 51% lower cost.
Benchmark Claims Against OpenAI
Anthropic's benchmark tables take direct aim at its rival. At default effort on FrontierCode, the company says Opus 5.5 beats OpenAI's GPT-6 Astra at roughly 20% of the cost per task. On Terminal-Bench 4.0 it matches Astra at about 40% of the cost, and on CursorBench it beats GPT-5.6 Sol by 11 points at about a third of the cost. Anthropic itself cautions that "benchmark margins have become a less reliable guide to real-world differences" at this capability level.
The Financial Times framed the release as Anthropic moving ahead of its expected IPO, while safety coverage noted the model's biology and cybersecurity capabilities are comparable to Claude Mythos 5.1, keeping it behind the same verification-program gates as Fable 5.1. Claude Sonnet 5.5 and Haiku 5.5 are promised in the coming weeks.
OpenAI's Answer: GPT-6 Sol and Luna
OpenAI's response landed minutes after Anthropic's announcement, according to Decrypt: GPT-6 Sol and GPT-6 Luna, both now generally available on Amazon Bedrock, as detailed in Amazon's announcement blog post. Amazon describes GPT-6 Astra as the family's upper tier for the most ambitious projects, with Sol positioned for demanding recurring work — implementing features, debugging, refactoring, code review, data analysis and multistep processes across tools — and Luna built for high-volume, focused tasks like document extraction, summarization and classification.
Amazon's post says both models carry "significantly lower API pricing than their GPT-5.6 predecessors," and The Next Web reported OpenAI cut GPT-6 prices in half with the release. On quality, Amazon relays an OpenAI internal evaluation finding that GPT-6 Sol makes approximately half as many factual mistakes as GPT-5.6 Sol, with improvements in coding and computer use. The Decoder, however, characterized the pair as cutting prices in half "but barely mov[ing] the needle on performance" — a reminder that this round of competition is about economics, not capability jumps.
From Slowdown Pact to Price War
The irony is hard to miss. Earlier this month, reporting on a slowdown agreement between the leading labs spawned an antitrust lawsuit naming Anthropic, OpenAI and others; Anthropic CEO Dario Amodei subsequently clarified his "pace the frontier" position as being about measurement, not moratoriums. Tuesday's releases are the first tangible evidence of what competition looks like when that pact is under legal threat.
The price war is also not new — OpenAI cut GPT-5.6 Sol API pricing by more than 20% earlier this month, and Chinese labs like DeepSeek have been aggressively undercutting on price all year. What is new is both US frontier labs moving at once, on the same day, at the same strategic layer: near-frontier quality at mid-tier prices.
Axios reported that cheap, powerful models are already increasing usage — a bullish signal for overall AI demand even as per-token revenue falls. For enterprises, the message is that frontier-adjacent work is getting dramatically cheaper; for the labs, the question is who can sustain margin at a price point that keeps falling.
---
Get the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
