Anthropic ships Claude Opus 4.7 with improved coding reliability and multimodal capabilities

Anthropic has released Claude Opus 4.7, maintaining its two-month release cadence established with recent Opus versions. The model targets advanced software engineering tasks and introduces several technical improvements alongside a tokenizer update that affects usage costs.

Enhanced coding capabilities with trade-offs

According to Anthropic, Opus 4.7 handles complex, long-running coding tasks with improved rigor and consistency compared to Opus 4.6. The company claims users can hand off difficult coding work that previously required close supervision with greater confidence. The model "devises ways to verify its own outputs before reporting back," according to the announcement.

Anthropic's benchmarks show Opus 4.7 outperforming Opus 4.6, GPT-5.4, and Gemini 3.1 Pro across agentic coding and computer use tasks. However, the model ranks behind Claude Mythos Preview, which remains unavailable to general users—Anthropic is only sharing it with key software platform vendors including Apple.

Tokenizer change increases usage costs

Opus 4.7 introduces an updated tokenizer that Anthropic says "improves how the model processes text." The trade-off: the same input maps to approximately 1.0-1.35× more tokens depending on content type. Additionally, the model "thinks more at higher effort levels, particularly on later turns in agentic settings," producing more output tokens. These changes directly impact token usage costs, though specific pricing was not disclosed in the announcement.

Multimodal and memory improvements

The model processes images at higher resolution than its predecessor. Anthropic claims it is "more tasteful and creative when completing professional tasks, producing higher-quality interfaces, slides, and docs."

Opus 4.7 includes improved file system-based memory functionality. According to Anthropic, the model "remembers important notes across long, multi-session work, and uses them to move on to new tasks that, as a result, need less up-front context."

Consistent release schedule established

Anthropic has maintained a two-month release cadence for its last three Opus versions: Opus 4.5, 4.6, and now 4.7. This follows a three-month gap between Opus 4.1 and 4.5, suggesting the company has settled into a more predictable upgrade schedule.

What this means

The tokenizer update represents a notable shift in Anthropic's approach—improving model performance while explicitly increasing token consumption by up to 35%. This trade-off forces users to balance enhanced capabilities against higher operational costs. The improved memory and verification features suggest Anthropic is positioning Opus 4.7 for autonomous, multi-session workflows where the model operates with less human oversight. The two-month cadence indicates Anthropic is prioritizing rapid iteration on software engineering capabilities, likely responding to competitive pressure from OpenAI's GPT-5 series and Google's Gemini models.