AiHubMix Documentation Hub

This article covers two key changes to reasoning control in Claude Opus 4.7, along with complete usage instructions for both the AIHubmix native API and the Chat unified interface. See also: Anthropic official announcement and model change log.

1. New Reasoning Control Features

✦ New `xhigh` Reasoning Effort Level

The new xhigh level sits between high and max, designed specifically for coding and Agentic tasks, striking a better balance between capability and efficiency.

low  ──  medium  ──  high  ──  xhigh ★NEW  ──  max

✦ Thinking Content Hidden by Default

In streaming responses, the thinking process is no longer shown by default. To receive a reasoning summary, explicitly pass the display field in your request:

display value	Opus 4.7	Opus 4.6	Behavior
”omitted”	Default	Non-default	Thinking block content is empty
”summarized”	Must set manually	Default	Returns thinking summary text

"reasoning": {
  "effort": "xhigh",
  "display": "summarized"
}

Figure: Opus 4.7 new xhigh level — Agentic coding performance comparison (Source: Anthropic official)

2. Claude Native API Reference

The `effort` values for the Anthropic native API are consistent with the official specification:

effort value	Supported Models	Description	Recommended Use Case
low	All supported models	Significant token savings with moderate capability trade-off	Simple tasks, high-concurrency requests, sub-agents
medium	All supported models	Balanced mode with moderate token savings	General agentic tasks
high	All supported models	Default; high capability performance	Complex reasoning, coding, agentic tasks
xhigh (new)	Opus 4.7 only	Extended capability between high and max; excels at long-horizon tasks	Recommended starting point for coding and agentic tasks
max	Opus series	Maximum capability	Frontier research problems

AIHubmix Claude Native API — Opus 4.7 Example

from anthropic import Anthropic

client = Anthropic(
    api_key="<AIHUBMIX_API_KEY>",
    base_url="https://aihubmix.com"
)

response = client.messages.create(
    model="claude-opus-4-7",
    max_tokens=16000,
    thinking={"type": "adaptive"},
    output_config={"effort": "xhigh"},  # Options: low / medium / high / xhigh / max
    messages=[
        {
            "role": "user",
            "content": "A snail is at the bottom of a 10-meter well. Each day it climbs 3 meters, but each night it slides back 2 meters. How many days does it take to reach the top?"
        }
    ]
)

print(response.content[-1].text)

3. Reasoning Effort Support in the AIHubmix Chat Unified Interface

The AIHubMix Chat unified interface aligns with the OpenAI specification and controls reasoning intensity via reasoning.effort. Different Claude models are automatically mapped to the corresponding effort level:

`reasoning_effort` (Reasoning Intensity Control) for Claude Models

effort value	Opus >=4.7 (new)	Opus 4.6 / 4.5	Sonnet 4.6
minimal	low	low	low
medium	medium	medium	medium
high	high	high	high
xhigh	xhigh	max	high
max	max	max	high

Note: xhigh is natively supported only on Opus 4.7. Other Opus models automatically fall back to max, and the Sonnet series falls back to high.

Chat Unified Interface — Opus 4.7 Example

from openai import OpenAI

client = OpenAI(
    base_url="https://aihubmix.com/v1",
    api_key="<AIHUBMIX_API_KEY>",
)

completion = client.chat.completions.create(
    model="claude-opus-4-7",
    # max_tokens=10000,  # Default is 4096; enable for longer outputs
    messages=[
        {
            "role": "user",
            "content": "A snail is at the bottom of a 10-meter well. Each day it climbs 3 meters, but each night it slides back 2 meters. How many days does it take to reach the top?"
        }
    ],
    extra_body={
        "reasoning": {"effort": "xhigh"}
    }
)

print(completion.choices[0].message.content)

4. Controlling Thinking Content in the Chat Unified Interface

In the OpenAI-compatible interface, reasoning supports a display field to control whether thinking summaries are returned. Claude Opus 4.7 does not return thinking content by default. Set "display": "summarized" to enable it.

Field	Value	Description
`display`	(omitted)	Thinking content not returned (default)
`display`	`"summarized"`	Returns a thinking summary

Opus 4.7 — Thinking Summary Example

from openai import OpenAI

client = OpenAI(
    base_url="https://aihubmix.com/v1",
    api_key="<AIHUBMIX_API_KEY>",
)

completion = client.chat.completions.create(
    model="claude-opus-4-7",
    messages=[
        {
            "role": "user",
            "content": "A snail is at the bottom of a 10-meter well. Each day it climbs 3 meters, but each night it slides back 2 meters. How many days does it take to reach the top?"
        }
    ],
    extra_body={
        "reasoning": {"effort": "xhigh", "display": "summarized"}
    }
)

print(completion.choices[0].message.content)

For further details, refer to the AIHubmix documentation or the Anthropic official docs

​1. New Reasoning Control Features

​✦ New xhigh Reasoning Effort Level

​✦ Thinking Content Hidden by Default

​2. Claude Native API Reference

​The effort values for the Anthropic native API are consistent with the official specification:

​AIHubmix Claude Native API — Opus 4.7 Example

​3. Reasoning Effort Support in the AIHubmix Chat Unified Interface

​reasoning_effort (Reasoning Intensity Control) for Claude Models

​Chat Unified Interface — Opus 4.7 Example

​4. Controlling Thinking Content in the Chat Unified Interface

​Opus 4.7 — Thinking Summary Example