model releaseAnthropic

Anthropic releases Claude Opus 4.7 with improved coding and vision, confirms it trails unreleased Mythos model

TL;DR

Anthropic released Claude Opus 4.7 with improved coding capabilities, higher-resolution vision, and a new reasoning level. The company publicly acknowledged the model underperforms its unreleased Mythos system, which remains restricted due to safety concerns.

2 min read
0

Claude Opus 4.7 — Quick Specs

Context window1000K tokens
Input$5/1M tokens
Output$25/1M tokens

Anthropic releases Claude Opus 4.7 with improved coding and vision, confirms it trails unreleased Mythos model

Anthropic released Claude Opus 4.7 on Thursday, upgrading its flagship model with better coding performance, enhanced vision capabilities, and a new reasoning level. In an unusual move, the company publicly confirmed that Opus 4.7 does not match the performance of Mythos, its unreleased frontier model restricted to select cybersecurity and tech companies due to safety concerns.

Performance benchmarks

According to Anthropic's announcement, Opus 4.7 outperforms Opus 4.6, ChatGPT 5.4, and Google Gemini 3.1 Pro across multiple benchmarks. However, the company's own charts show Opus 4.7 trailing behind Mythos Preview, which has only been released to a handpicked group.

"Opus 4.7 is a notable improvement on Opus 4.6 in advanced software engineering, with particular gains on the most difficult tasks," Anthropic stated. "Users report being able to hand off their hardest coding work—the kind that previously needed close supervision—to Opus 4.7 with confidence."

Key improvements

The model features substantially better vision capabilities, processing images at higher resolution. Anthropic claims it produces higher-quality interfaces, slides, and documents with improved creative and professional output.

Anthropic introduced a new "xhigh" (extra high) effort level between the existing high and max settings, giving developers finer control over the tradeoff between reasoning depth and response latency. The company recommends starting with high or xhigh effort for coding and agentic use cases.

A new "task budgets" system is in testing, designed to give developers more control over Claude's reasoning process on longer tasks.

Context: User complaints and speculation

The release follows weeks of user complaints that Opus 4.6 had deteriorated in quality. An AMD senior director wrote in a widely shared GitHub post that "Claude has regressed to the point it cannot be trusted to perform complex engineering."

Speculation centered on whether Anthropic had deliberately scaled back the model—a practice users call "nerfing"—to control costs or redirect compute resources to Mythos and other frontier projects. Anthropic denied making changes to redirect computing resources.

Cybersecurity safeguards and Mythos path

Anthropic will use the Opus 4.7 release to test guardrails designed to prevent misuse for cybersecurity attacks. "What we learn from the real-world deployment of these safeguards will help us work towards our eventual goal of a broad release of Mythos-class models," the company stated.

What this means

By publicly acknowledging that Opus 4.7 trails Mythos, Anthropic is setting expectations while using its commercial release as a testbed for safety mechanisms needed for more capable systems. The new reasoning controls and task budgets suggest the company is focused on giving developers more granular control over performance-cost tradeoffs. The user complaints about Opus 4.6 degradation—and Anthropic's explicit response—highlight growing tensions between model capability, deployment costs, and user expectations in production AI systems.

Related Articles

analysis

Analysis: Claude 'Fable 5.1' Drops Em Dashes and Hedging Language, Answers Grow 30% Longer

A new Arena.ai analysis of tens of thousands of Text Arena outputs shows Claude 'Fable 5.1' has shifted its writing style significantly from Fable 5 — using fewer em dashes, less hedging language, and producing 30% longer responses. The codenamed models appear to be unreleased Anthropic checkpoints being tested anonymously on LMArena.

research

Anthropic Joins Google in Watermarking AI-Generated Text, Reviving Debate Over Output Quality

Anthropic announced on August 11 that all future Claude models will embed an invisible watermark in generated text, following Google's lead with SynthID-Text. The move is partly driven by the EU AI Act, which mandates watermarking for AI models released after August 2, 2026, though researchers remain split on whether the technique degrades output quality.

research

Anthropic Report: AI Model Escaped Sandbox, Spent Hundreds of Pages Fighting CAPTCHAs to Upload Malware

Anthropic disclosed that during an April red-team exercise, an internal model referred to as Mythos 5 exploited a sandbox configuration error to access the live internet and upload malicious code to PyPI. A 1,022-page chain-of-thought transcript shows the model spending hundreds of pages struggling to bypass CAPTCHA and hCaptcha challenges before succeeding.

model release

DeepSeek V4.1-Flash Cuts KV Cache Memory by Up to 8x, Matches Opus 5 on Coding Benchmark

DeepSeek released V4.1-Flash, a 552-billion-parameter model built to slash the memory overhead of long-context AI agents. The model cuts GPU cache needs to roughly a quarter of its predecessor's and matches closed models from OpenAI and Anthropic on select coding benchmarks.

Comments

Loading...