anthropic/claude-opus-5
The July 2026 Opus — succeeded by Opus 5.5.
Tune the request, see the current price ceiling, and inspect the real response without leaving this model page.
Sign in first to upload images to this vision model.
One API key, one interface, and a single command to call models from OpenAI, Anthropic, Google, and a few dozen others. Swap providers by changing a string, not your code, and compare outputs side by side when you're deciding which model fits. Requests, token counts, and per-call costs are logged locally so you can see where your spend is going.
Claude Opus 5 was released by Anthropic on July 24, 2026, succeeding Opus 4.8. It became legacy when Opus 5.5 arrived on September 22, 2026, with retirement not sooner than July 24, 2027. It has a 1M-token context window (both default and maximum), up to 128K output tokens per request, image input and tool calling, and a reliable knowledge cutoff of May 2026. At launch Anthropic called it a step-change improvement over Opus 4.8, much stronger at verifying its own work and iterating until it succeeds, and close to the frontier intelligence of Fable 5.
Opus 5 was the first Opus with thinking on by default, but it kept a switch: at high effort or below you can set thinking to "disabled"; at xhigh or max that returns a 400. All five effort levels are available, the API default is high, and the minimum cacheable prompt is 512 tokens. Anthropic notes that on Opus 5 effort changes how much the model thinks, not reliably how long it answers, so length should be controlled in the prompt.
Opus 5.5 changed several things that now fail outright: thinking cannot be disabled; forced tool use (tool_choice "any" or a named tool) returns a 400; the computer_20251124 computer use tool is no longer accepted on the Claude API; the default effort drops to medium; and text written between tool calls moves into thinking blocks that are empty by default. An integration that depends on any of these (turning thinking off to save tokens, forcing a tool call, streaming progress text between tool calls) can stay on Opus 5 without code changes. Otherwise, Anthropic recommends Opus 5.5.
Through RouteMux, Opus 5 costs $1.4 per 1M input tokens against Anthropic's $5, $7 per 1M output tokens against $25, and $0.14 per 1M cache-read tokens against $0.5.
Prices verified 2026-10-06
Cache read $0.14/M · Cache write $1.75/M
"disabled" is accepted at high effort or below. From Opus 5.5 it cannot be turned off at any level.tool_choice "any" or a named tool is honored, which suits older code that forces a call to get structured output. Opus 5.5 and Fable 5.1 return a 400.Limited time10% extra Credits on your first top-up — up to 200 bonus Credits
Applies to your first top-up within 72 hours of signing up. Minimum $20, once per account.
Use any of these IDs to call this model via the API.
anthropic/claude-opus-5Replace the ROUTEMUX_KEY placeholder with your API key. Create one →
from anthropic import Anthropic
client = Anthropic(
base_url="https://api.routemux.com/anthropic",
api_key="<ROUTEMUX_KEY>",
)
message = client.messages.create(
model="anthropic/claude-opus-5",
max_tokens=1024,
messages=[{"role": "user", "content": "What is the meaning of life?"}],
)
print(message.content[0].text)The July 2026 Opus — succeeded by Opus 5.5.
Claude Opus 5 supports a context window of up to 1M tokens, with up to 128K output tokens per request.
Through RouteMux, Claude Opus 5 costs $1.4 for input and $7 for output (USD per 1M tokens), below the official list price.
Claude Opus 5 is OpenAI-compatible: point your base URL at RouteMux and set the model field to anthropic/claude-opus-5 — no code changes needed.
Claude Opus 5 accepts text, image as input and produces text as output.
Claude Opus 5 was released on July 24, 2026.
Reviewed by the RouteMux team on 2026-10-24
2,409 requests · last 30 days