Extended Variant

Not yet available.

Some models offer an extended context window beyond their standard limit. Append :extended to an eligible model slug to unlock that larger window. This is useful for processing long documents, large codebases, or multi-turn conversations that exceed the default context length.

Usage#

curl https://opendunes.com/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENDUNES_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-opus-4.8:extended",
    "messages": [
      { "role": "user", "content": "Summarize this 200-page document: ..." }
    ]
  }'

The :extended suffix routes the request to the provider and model configuration that supports the larger context. The model field in the response echoes the full slug with the suffix.

Extended vs. standard context#

The exact context limits for each model are in the catalog at GET /api/models/{slug}. The :extended context size is listed as context_length_extended when the model supports it.

When to use extended context#

Extended context is best for:

• Document analysis — contracts, reports, or transcripts that overflow the standard window
• Codebase-aware generation — passing large amounts of code in a single prompt
• Long conversation history — keeping more turns in context without truncation

For most interactive use cases, the standard context window is sufficient and costs less per token in DA.

Pricing (DA)#

Extended-context requests are billed at the model's :extended per-token rate, which is higher than the standard rate. The exact DA cost per million input/output tokens is in the model catalog. Because prompts using extended context tend to be longer, estimate your request cost before sending to avoid surprises.

Use the catalog endpoint to compare rates:

GET /api/models/anthropic/claude-opus-4.8

The response includes both pricing.standard and pricing.extended in DA per million tokens.

Using with the OpenAI SDK#

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://opendunes.com/api/v1",
    api_key=os.environ["OPENDUNES_API_KEY"],
)

resp = client.chat.completions.create(
    model="anthropic/claude-opus-4.8:extended",
    messages=[{"role": "user", "content": "Analyze the following legal text..."}],
    max_tokens=4096,
)
print(resp.choices[0].message.content)

Note. Requesting :extended on a model that does not support it returns a 404 model_not_found error. Confirm support in the catalog before using this variant in production.