Anthropic SDK v0.120.0 Adds Reference to Unannounced 'Claude Opus 5' Model
The anthropic-sdk-python v0.120.0 release adds a reference to a model identifier called claude-opus-5, the first public sign of a next-generation Opus model. Anthropic has not issued an official announcement, and no pricing, context window, or benchmark data has been disclosed.
SDK Changelog Reveals Model Identifier, No Official Announcement Yet
Anthropic shipped version 0.120.0 of its official Python SDK (anthropic-sdk-python) on July 24, 2026, and the release notes contain the first public reference to a model identifier named claude-opus-5. The mention appears in a single commit alongside two unrelated API additions, with no accompanying documentation, pricing, or specification changes.
This is a changelog entry, not a product launch. Anthropic has not published a model card, blog post, or pricing page for Claude Opus 5 as of this writing. The identifier's presence in the SDK typically precedes a public release by days to weeks, based on Anthropic's historical pattern of shipping client-side support ahead of announcements, but the company has made no confirmed statement about availability, capabilities, or release timing.
What actually shipped in v0.120.0
Three features are listed in the full changelog (diff: v0.119.0...v0.120.0):
claude-opus-5model identifier added to the API client — enables the SDK to recognize and route requests to the model, but contains no metadata about context window, modality support, or benchmark performance.- Tool addition/removal blocks and
tool_changeevents — new event types that let developers add or remove tools mid-conversation and receive structured notifications when tool availability changes during a session. This is a functional SDK capability independent of any specific model. - Expanded client-side fallback credit token types and a server-side fallbacks default option — broadens how the SDK handles fallback billing/credit accounting when primary model routes are unavailable, with a new default behavior for server-managed fallbacks.
All three changes are confirmed by the commit history at the linked GitHub release. Beyond the identifier string itself, Anthropic has disclosed zero specifications for Claude Opus 5: no context window size, no input/output pricing per million tokens, no benchmark scores (MMLU, HumanEval, or otherwise), and no confirmed training cutoff date.
What this means
SDK-level model identifiers are a standard early signal in Anthropic's release cadence — Claude 3.5 Sonnet, Claude 4, and Claude 4.1 all appeared in SDK code shortly before public launch. Their presence here is a credible indicator that Anthropic is preparing to ship a successor to Claude Opus 4.1, but it is not confirmation of a release date or feature set. Developers should not build production integrations against claude-opus-5 yet, since the identifier could change, be gated behind an internal flag, or be pulled before general availability.
The more immediately useful changes for developers are the tool_change events and expanded fallback credit handling, both of which are live now and affect any application using dynamic tool calling or fallback routing across Anthropic's model lineup. Anthropic has not responded to requests for comment on Claude Opus 5's timeline; this article will be updated when the company issues an official announcement.
Related Articles
Anthropic's Claude Fable 5.1 Reportedly Solves 1653 Royalist Cipher in 44 Minutes
According to testing firm Vals AI, Anthropic's Claude Fable 5.1 independently identified and solved the 'Cyphral Distich,' a 1653 numeric cipher by Sir Thomas Urquhart that had defeated other frontier models. The AI decoded a hidden pro-royalist message by mapping each number to a word in Urquhart's original text.
Alibaba Releases Qwen-Drive 1.0, an Open Driving Model That Explains Its Own Decisions
Alibaba has released Qwen-Drive 1.0, a driving model built on Qwen3.5-4B that handles spatial perception, route planning, and cockpit dialogue in a single system. Reinforcement learning cut the rate of off-road driving errors in simulation from 24 percent to 12 percent, though the model's stated reasoning doesn't always match its actual maneuvers.
Microsoft Releases VibeVoice-ASR-Streaming-7B, an Open-Weight Streaming Speech Recognition Model with Speaker Attributio
Microsoft Research has released VibeVoice-ASR-Streaming-7B, an open-weight streaming automatic speech recognition model that transcribes both who is speaking and what they say in real time. The model, listed at 9B parameters despite its name, supports 10 languages and custom hotwords under an MIT license.
Google's WeatherNext 3 Drops Physics Simulations, Learns Weather Forecasting Directly From Satellite Data
Google and DeepMind released WeatherNext 3, an AI weather model that trains directly on live geostationary satellite data instead of physics-based simulations. The model produces hourly forecasts at up to 5-kilometer resolution and now powers weather features in Google Search, Maps, and Gemini.
Comments
Loading...