Add OpenAI GPT-Image-2.5 (Flare, Sunburst) to the model registry - #162
Merged
Merged
Conversation
Registers both variants of OpenAI's GPT-Image-2.5 family, GA 2026-09-08: gpt-image-2.5-flare (fast, general-purpose, default alias) and gpt-image-2.5-sunburst (highest-precision editing). Both route through /images/generations and /images/edits exactly like gpt-image-2, and per OpenAI's model pages "token rates align with GPT Image 2", so they reuse the same flat $0.05/image approximation. Companion PR: opengradient/chat-api adds the catalog entries. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DoBWfeNKzcQ2ZscrGeUQsA
adambalogh
marked this pull request as ready for review
September 10, 2026 12:36
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
OpenAI shipped GPT-Image-2.5 on 2026-09-08 (two days before this run) as
two GA API models:
gpt-image-2.5-flare— "our fastest model for high-quality, everydayimage generation": 50% lower latency than
gpt-image-2, betterreference-photo fidelity and precise edits. Positioned as the general-purpose
successor to
gpt-image-2. Registered with a baregpt-image-2.5aliassince it's the default/general-purpose variant.
gpt-image-2.5-sunburst— "our most capable model for image generationand editing": precision editing for premium/production visual workflows.
Both route through the same
/images/generations+/images/editsendpointsas
gpt-image-2(not chat/Responses/Batch), and per OpenAI's own model pages,"token rates align with GPT Image 2" ($5/$1.25(cached) per MTok text,
$8/$2(cached) per MTok image input, $30 per MTok image output — text output
unbilled) — so both reuse the same flat
$0.05/image approximationgpt-image-2already uses.Pricing / parameter citations
gpt-image-2.5-flare,
gpt-image-2.5-sunburst.
/v1/images/generationsand/v1/images/editsare supported endpoints for both variants (confirmed explicitly for
Sunburst: "Cannot be used via Chat Completions, Responses, or Batch
endpoints"), matching
gpt-image-2's existing request shaping(
image_response_format=None,image_supports_reference=True,image_edit_endpoint="/images/edits"), so no new flags were needed.Found during
Routine check for new releases from providers already integrated in the
gateway. Also checked the two other open PRs touching
model_registry.pyfrom previous runs of this task — #160 ("Add Tencent Hy4 Preview") and #156
("Add Z.ai GLM-5.3 and GLM-5.3-Flash") — both are unrelated (different
provider sections, no line overlap) and still open drafts, so no changes were
bundled into them.
Also checked Claude Mythos 5.1 and Gemini 3.8 Flash Cyber, both released
around the same time — neither was added: both are vetted-access-only
cybersecurity/biology models (Anthropic's Project Glasswing, Google's
Fairwind Program) with no public self-serve API, so there's nothing a
standard provider API key here could route to.
Test plan
uv run pytest tests/test_pricing.py— 147 passed (addedtest_gpt_image_2_5_flare_resolves,test_gpt_image_2_5_bare_alias_resolves_to_flare,test_gpt_image_2_5_sunburst_resolves)make lint(ruff format + check, mypy) — cleanCompanion PR
opengradient/chat-api#299 — adds the corresponding catalog entries so clients
can select the model.
🤖 Generated with Claude Code
https://claude.ai/code/session_01DoBWfeNKzcQ2ZscrGeUQsA