Anthropic Ships Claude Opus 5.5, OpenAI Launches GPT-6 Sol and Luna — All Cheaper Than Predecessors
Anthropic released Claude Opus 5.5 at $4/$20 per million input/output tokens, undercutting Opus 5's $5/$25 pricing while claiming better agentic coding scores. OpenAI countered with GPT-6 Sol ($2/$10) and GPT-6 Luna ($0.10/$0.50), both up to 50% cheaper than GPT-5.6's promotional rates.
Anthropic and OpenAI cut prices while claiming performance gains
Anthropic and OpenAI both released new models this week, each pitched as cheaper and more capable than their immediate predecessors — a move that runs counter to recent industry talk about "pacing the frontier" of AI development.
Claude Opus 5.5
Anthropic's Opus 5.5 replaces Opus 5 and is priced at $4 per million input tokens and $20 per million output tokens, down from $5 and $25 respectively. Anthropic says the model is better at "finding and fixing inefficiencies in software," financial analysis, and general business work — capabilities aimed squarely at enterprise customers.
According to Anthropic, Opus 5.5 outperforms OpenAI's GPT-6 Astra on two agentic coding benchmarks: Terminal-Bench 4.0 and FrontierCode v1.1 (Main). The company did not publish the specific scores. Anthropic also claims the model produces clearer writing and "attempted to circumvent boundaries around 85 percent less often" than earlier Claude models — a claim based on Anthropic's own internal testing, not independently verified.
Opus 5.5 is available now through the Claude platform, Amazon Web Services, Google Cloud, and Microsoft Azure.
GPT-6 Sol and GPT-6 Luna
OpenAI released two new models trained with methods similar to GPT-6 Astra: GPT-6 Sol and GPT-6 Luna. Both are priced up to 50 percent below the promotional rates of their GPT-5.6 counterparts. GPT-6 Sol costs $2 per million input tokens and $10 per million output tokens. GPT-6 Luna, positioned as the lower-cost option, costs $0.10 per million input tokens and $0.50 per million output tokens.
OpenAI claims GPT-6 Sol makes roughly half as many factual errors as its predecessor and matches the coding performance of Anthropic's Fable 5.1 model at a lower price point — another unverified company claim rather than a third-party benchmark result. Both new models reportedly communicate more clearly and, thanks to improved prompt caching, reuse more context to respond faster, according to OpenAI.
On alignment, OpenAI says the GPT-6 models lie less often about the results of coding tasks and reject more attempts to bypass safety restrictions when given unsafe commands, based on the company's internal testing.
GPT-6 Sol and GPT-6 Luna are live now in ChatGPT Work and Codex for Plus, Pro, Business, and Enterprise subscribers. Free users and Go subscribers get access only to GPT-6 Luna through OpenAI's desktop app.
Industry standards push
Both companies have proposed their own criteria for third-party evaluators to assess AI progress and safety. Neither Anthropic nor OpenAI has said whether these proposals will converge into a shared industry standard, and no external body has adopted either framework yet.
What this means
This release cycle signals that price-per-token, not just raw capability, has become the primary competitive battleground between Anthropic and OpenAI. Both companies are shipping models that claim similar or better performance than their predecessors at 20-50 percent lower cost — a pattern that will pressure smaller labs and open-weight providers to match on price rather than compete purely on benchmarks. Notably, nearly every performance claim in both announcements comes from internal testing rather than independent verification, including the head-to-head comparisons between Opus 5.5 and GPT-6 Astra. Buyers evaluating these models for production use should wait for third-party benchmark results before treating the vendor comparisons as settled.
Related Articles
Anthropic tells Claude Opus 5.5 users to drop "think hard" prompts and start at Medium effort
Anthropic says users should remove phrases like "think hard" from prompts for Claude Opus 5.5, which reasons adaptively on its own. The company also recommends starting at the default Medium effort level, which it says matches Opus 5 on High in internal tests.
Claude can now build HTML-in-Canvas sites, needs Chrome or Edge 157 or Claude desktop 5.11492
Claude can now build web pages that embed real HTML elements such as text and buttons inside canvas graphics and 3D effects. The feature works in Chrome and Edge from version 157 and in Claude's desktop apps from version 5.11492. Users must request HTML-in-Canvas explicitly in their prompt.
Anthropic: Selective notes beat full logs for Claude agent memory, with 'Dreaming' fixing errors
Anthropic published research on how Claude retains knowledge across sessions using note files. Its tests show selective notes outperform exhaustive archives, and a post-session process called Dreaming can correct faulty notes. All results come from Anthropic's own tests.
Anthropic says Claude Mythos Preview onboarded AMD MI355X GPUs in about three days, versus six to 12 months
Anthropic told IPO investors that Claude Mythos Preview did the software work to run its AI on AMD MI355X GPUs in about three days, a job it says historically took six to 12 months. The three days cover prompt processing for one model only, and no performance comparison with Nvidia has been given.
Comments
Loading...