Skip to content

feat(bedrock): add openai.gpt-daybreak-blue-5.6-sol - #1007

Merged
naresh4dev merged 1 commit into
mainfrom
feat/bedrock-daybreak-blue-5.6-sol
Oct 28, 2026
Merged

naresh4dev merged 1 commit into
mainfrom
feat/bedrock-daybreak-blue-5.6-sol

Conversation

@naresh4dev

Copy link
Copy Markdown
Member

What

Adds openai.gpt-daybreak-blue-5.6-sol to the Amazon Bedrock bedrock-mantle provider (config + pricing).

Why

New model available on AWS Bedrock (OpenAI's Daybreak Blue: GPT-5.6 Sol, for authorized cyber-defense use cases). Only exposed via the bedrock-mantle endpoint (Chat Completions / Responses APIs), not bedrock-runtime.

How

  • general/bedrock-mantle.json: config entry mirrors sibling openai.gpt-5.6-sol — chat, tools+image support, max_completion_tokens up to 272000, legacy sampling params removed (same reasoning-model shape as the rest of the openai.gpt-5.6-*/gpt-6.* family on bedrock-mantle).
  • pricing/bedrock-mantle.json: pricing entry, short-context/in-region tier, converted from $/1M to cents-per-token per repo convention (price = dollars_per_million / 10000):
Field $ / 1M tokens repo price
Input $4.40 0.00044
Input cache write (30-min) $5.50 0.00055
Input cache read $0.44 0.000044
Output $22.00 0.0022

Note: AWS also publishes a long-context tier (>272K input tokens, 2x price). bedrock-mantle.json has no context_tiers wrapper precedent for any sibling model in this file (that schema is only used in pricing/bedrock.json, the classic Converse/Invoke endpoint, which this model doesn't support), so this PR keeps the flat short-context pricing shape to match existing openai.gpt-5.6-*/gpt-6.* bedrock-mantle entries. Happy to add a context-tiered variant if desired.

Testing

  • jq empty on both changed files — valid JSON
  • python3 scripts/check_duplicate_keys.py — no duplicate keys across the repo
  • npx eslint/prettier via lint-staged pre-commit hook — clean

Source

https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-openai-gpt-daybreak-blue-56-sol.html

- Config mirrors sibling openai.gpt-5.6-sol (chat, tools+image,
  max_completion_tokens up to 272000)
- Pricing (short-context, in-region tier, USD/1M -> cents-per-token):
  input 4.40 -> 0.00044, cache-write 5.50 -> 0.00055,
  cache-read 0.44 -> 0.000044, output 22.00 -> 0.0022

Source: https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-openai-gpt-daybreak-blue-56-sol.html
Copilot AI balanced review requested due to automatic review settings October 28, 2026 15:55

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟡 Changes recommended

The pricing entry omits the 2× long-context tier, causing requests above 272K input tokens to be underpriced because the output-token limit does not constrain input.

1 open issue
What changed in this PR

Adds openai.gpt-daybreak-blue-5.6-sol to the Amazon Bedrock Mantle model catalog, with model capabilities and pricing for the Mantle endpoint.

Changes:

  • Configures chat, tool, and image support with reasoning-model parameter restrictions.
  • Adds short-context token, cache, and search pricing.
File Description
pricing/​bedrock-mantle.json Adds token, cache, and search rates for the new model.
general/​bedrock-mantle.json Registers model capabilities and completion-token limits.

Review effort: Balanced

Comment thread pricing/bedrock-mantle.json Resolved
@naresh4dev
naresh4dev merged commit 8872138 into main Oct 28, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants