feat(openai): add gpt-5.5 models by smakosh · Pull Request #2076 · theopenco/llmgateway

smakosh · 2026-04-24T20:22:39Z

Summary

This PR adds support for two new OpenAI models: GPT-5.5 and GPT-5.5 Pro, with configurations for both OpenAI and Azure providers.

Key Changes

GPT-5.5: Added base model configuration with 1.05M context size, 128K max output, vision, tools, web search, and reasoning capabilities
- OpenAI provider: $5.00/$30.00 per 1M input/output tokens
- Azure provider: Same pricing with 30% discount and marked as unstable
GPT-5.5 Pro: Added premium variant with enhanced reasoning using more compute
- OpenAI provider: $30.00/$180.00 per 1M input/output tokens with 30% discount
- Azure provider: Same pricing without discount
- Both providers marked with test: "skip" flag
- JSON output schema support disabled (vs enabled in base model)

Notable Implementation Details

Both models released on 2026-04-23
Both support streaming, vision, tools, web search ($0.01 per search), and reasoning with "omit" output mode
Both support Responses API and JSON output
Context window: 1,050,000 tokens with 128,000 token max output
Cached input pricing available for GPT-5.5 ($0.5 per 1M tokens)
Standard parameters supported: temperature, top_p, frequency_penalty, presence_penalty, response_format

https://claude.ai/code/session_01XrSSeRjBSUYtNLurCbG9Nm

Summary by CodeRabbit

New Features
- Added two new GPT models: gpt-5.5 and gpt-5.5-pro for advanced AI use cases.
- Both models enable enhanced reasoning and JSON output; gpt-5.5 includes JSON schema support while gpt-5.5-pro does not.
- Platform-optimized provider configurations (pricing, context/output capacities, streaming, vision, tools, web search support) included to support diverse operational needs.

coderabbitai · 2026-04-24T20:22:56Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: f06e472a-f46c-4a79-aa17-86e67ce773b7

📥 Commits

Reviewing files that changed from the base of the PR and between 97bb9af and e55125c.

📒 Files selected for processing (1)

packages/models/src/models/openai.ts

Walkthrough

Adds two new model entries, gpt-5.5 and gpt-5.5-pro, to the exported openaiModels array with provider-specific configurations for openai and azure (pricing, contexts, capability flags, Responses API support, and JSON output settings).

Changes

Cohort / File(s)	Summary
GPT-5.5 Model Definitions `packages/models/src/models/openai.ts`	Added two exported model definitions (`gpt-5.5`, `gpt-5.5-pro`) with dual provider blocks (`openai`, `azure`). Includes pricing (input/output/cached, discounts), context and max token settings, capability flags (`streaming`, `vision`, `tools`, `webSearch` + `webSearchPrice`), `reasoning` fields, `supportsResponsesApi: true`, and `jsonOutput`/`jsonOutputSchema` configuration. `gpt-5.5` declares supported params; `gpt-5.5-pro` providers marked `test: "skip"`.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Possibly related PRs

feat(models): add gpt-image-2 from openai #2059 — Modifies the same openaiModels array in packages/models/src/models/openai.ts, adding new OpenAI model entries.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'feat(openai): add gpt-5.5 models' accurately summarizes the main change: adding support for GPT-5.5 model variants to the OpenAI models configuration.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

Create stacked PR
Commit on current branch

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch claude/add-gpt-5.5-models-QS0Sd

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

coderabbitai

🧹 Nitpick comments (1)

packages/models/src/models/openai.ts (1)
1539-1655: Optional: consider placing the new entries near other gpt-5.x siblings.

gpt-5.5 / gpt-5.5-pro are inserted between gpt-5.4-nano and gpt-5.2-codex, which further fragments the already-mixed ordering of the gpt-5.x family in this array. Grouping them after gpt-5.4-nano alongside the rest of the 5.x series (or consistently by release date) would make the file easier to scan. Purely cosmetic; no behavioral impact.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.

In `@packages/models/src/models/openai.ts` around lines 1539 - 1655, The new model
entries with id "gpt-5.5" and "gpt-5.5-pro" are placed between "gpt-5.4-nano"
and "gpt-5.2-codex", breaking the logical grouping of the gpt-5.x family;
relocate the entire objects for gpt-5.5 and gpt-5.5-pro so they appear adjacent
to the other gpt-5.x siblings (for example, immediately after the "gpt-5.4-nano"
entry) to keep the 5.x models grouped consistently in the array.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@packages/models/src/models/openai.ts`:
- Around line 1539-1655: The new model entries with id "gpt-5.5" and
"gpt-5.5-pro" are placed between "gpt-5.4-nano" and "gpt-5.2-codex", breaking
the logical grouping of the gpt-5.x family; relocate the entire objects for
gpt-5.5 and gpt-5.5-pro so they appear adjacent to the other gpt-5.x siblings
(for example, immediately after the "gpt-5.4-nano" entry) to keep the 5.x models
grouped consistently in the array.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 0fbd7db7-09df-468e-9e21-8e483b69f1b8

📥 Commits

Reviewing files that changed from the base of the PR and between fac1a7b and 577a975.

📒 Files selected for processing (1)

packages/models/src/models/openai.ts

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

## Summary OpenAI's API has no numeric reasoning-budget primitive — only `reasoning.effort` (none/minimal/low/medium/high/xhigh). The `reasoningMaxTokens: true` flag on `gpt-5.5` and `gpt-5.5-pro` (added in #2076) was unsupported upstream: a request with `reasoning.max_tokens: 1024` is rejected by OpenAI with `Unknown parameter: 'reasoning.max_tokens'`. The flag is dropped from those provider mappings; the gateway now correctly returns 400 instead of silently sending a budget that gets stripped or rejected. The earlier e2e `test.each(reasoningMaxTokensModels)` round-tripped a real completion against every capable model just to inspect the upstream request body — wasteful (real reasoning tokens are not free) and the response-shape assertions duplicated `basic reasoning`. Replaced with focused unit tests on `prepareRequestBody` covering each provider that maps the budget: - `anthropic` → `thinking.budget_tokens` - `aws-bedrock` → `additionalModelRequestFields.thinking.budget_tokens` - `google-ai-studio` → `generationConfig.thinkingConfig.thinkingBudget` - `google-vertex` → `generationConfig.thinkingConfig.thinkingBudget` The negative e2e in `api-individual.e2e.ts` (gateway returns 400 when `reasoning.max_tokens` is sent to a non-capable model) is kept — it short-circuits at validation, no provider call. ## Test plan - [x] `pnpm vitest run packages/actions/src/prepare-request-body.spec.ts` — 35/35 pass, including 4 new forwarding tests. - [x] `pnpm vitest run -c vitest/vitest.e2e.config.mts apps/gateway/src/api-individual.e2e.ts -t "reasoning.max_tokens error"` — gateway 400 path still verified. - [x] `pnpm format` clean; full build succeeds. 🤖 Generated with [Claude Code](https://claude.com/claude-code)  ## Summary by CodeRabbit * **Bug Fixes** * Corrected reasoning.max_tokens support status for gpt-5.5 and gpt-5.5-pro models on OpenAI and Azure. * Enhanced validation to reject reasoning.max_tokens requests on models that don't support this feature with appropriate error messaging. * **Tests** * Added end-to-end tests for reasoning.max_tokens handling and validation across providers.  --------- Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>

…nco#2079) ## Summary OpenAI's API has no numeric reasoning-budget primitive — only `reasoning.effort` (none/minimal/low/medium/high/xhigh). The `reasoningMaxTokens: true` flag on `gpt-5.5` and `gpt-5.5-pro` (added in theopenco#2076) was unsupported upstream: a request with `reasoning.max_tokens: 1024` is rejected by OpenAI with `Unknown parameter: 'reasoning.max_tokens'`. The flag is dropped from those provider mappings; the gateway now correctly returns 400 instead of silently sending a budget that gets stripped or rejected. The earlier e2e `test.each(reasoningMaxTokensModels)` round-tripped a real completion against every capable model just to inspect the upstream request body — wasteful (real reasoning tokens are not free) and the response-shape assertions duplicated `basic reasoning`. Replaced with focused unit tests on `prepareRequestBody` covering each provider that maps the budget: - `anthropic` → `thinking.budget_tokens` - `aws-bedrock` → `additionalModelRequestFields.thinking.budget_tokens` - `google-ai-studio` → `generationConfig.thinkingConfig.thinkingBudget` - `google-vertex` → `generationConfig.thinkingConfig.thinkingBudget` The negative e2e in `api-individual.e2e.ts` (gateway returns 400 when `reasoning.max_tokens` is sent to a non-capable model) is kept — it short-circuits at validation, no provider call. ## Test plan - [x] `pnpm vitest run packages/actions/src/prepare-request-body.spec.ts` — 35/35 pass, including 4 new forwarding tests. - [x] `pnpm vitest run -c vitest/vitest.e2e.config.mts apps/gateway/src/api-individual.e2e.ts -t "reasoning.max_tokens error"` — gateway 400 path still verified. - [x] `pnpm format` clean; full build succeeds. 🤖 Generated with [Claude Code](https://claude.com/claude-code)  ## Summary by CodeRabbit * **Bug Fixes** * Corrected reasoning.max_tokens support status for gpt-5.5 and gpt-5.5-pro models on OpenAI and Azure. * Enhanced validation to reject reasoning.max_tokens requests on models that don't support this feature with appropriate error messaging. * **Tests** * Added end-to-end tests for reasoning.max_tokens handling and validation across providers.  --------- Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>

feat: add gpt-5.5 and gpt-5.5-pro openai models

577a975

github-actions Bot changed the title ~~Add GPT-5.5 and GPT-5.5 Pro model configurations~~ feat(openai): add gpt-5.5 models Apr 24, 2026

smakosh self-assigned this Apr 24, 2026

coderabbitai Bot reviewed Apr 24, 2026

View reviewed changes

steebchen approved these changes Apr 25, 2026

View reviewed changes

smakosh and others added 3 commits April 25, 2026 12:16

Merge branch 'main' into claude/add-gpt-5.5-models-QS0Sd

52b8681

test: skip e2e for gpt-5.5 openai provider

97bb9af

feat(models): enable reasoningMaxTokens for gpt-5.5

e55125c

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

smakosh added this pull request to the merge queue Apr 25, 2026

Merged via the queue into main with commit 62f7174 Apr 25, 2026
17 of 18 checks passed

smakosh deleted the claude/add-gpt-5.5-models-QS0Sd branch April 25, 2026 11:09

This was referenced Apr 27, 2026

fix(azure): unmark unstable, skip e2e tests #2100

Merged

feat(azure): add gpt-oss-120b #2101

Merged

steebchen mentioned this pull request May 7, 2026

test(reasoning): unit-test budget forwarding; fix openai flag #2079

Merged

3 tasks done

coderabbitai Bot mentioned this pull request May 12, 2026

feat(gateway): add OpenAI-compatible embeddings #2280

Merged

This was referenced May 31, 2026

feat(providers): add openai-discount provider #2471

Closed

feat: support OpenAI service tiers #2568

Closed

fix: remove hardcoded discounts #2554

Merged

coderabbitai Bot mentioned this pull request Jun 7, 2026

feat: add provider service tiers #2572

Merged

This was referenced Jul 9, 2026

feat(models): add OpenAI GPT-5.6 sol/terra/luna #2970

Merged

feat(models): enable verbosity on older GPT-5 models #2989

Merged

Navigation Menu

feat(openai): add gpt-5.5 models - #2076

feat(openai): add gpt-5.5 models#2076
smakosh merged 4 commits into
mainfrom
claude/add-gpt-5.5-models-QS0Sd

smakosh commented Apr 24, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented Apr 24, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

smakosh commented Apr 24, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Key Changes

Notable Implementation Details

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Apr 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Estimated code review effort

Possibly related PRs

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

smakosh commented Apr 24, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Apr 24, 2026 •

edited

Loading