feat(image-generator): add gpt-image-2 model support

[waleedlatif1]

Summary

• Add gpt-image-2 as a model option in the Image Generator block
• Wire up gpt-image-2 specific params: size (auto, 1024², 1536×1024, 1024×1536, 2048², 3840×2160), quality (auto/low/medium/high), background (auto/opaque — transparent not supported per docs), output_format (png/jpeg/webp), moderation (auto/low)
• Share quality/background/output_format wiring with gpt-image-1; moderation is gpt-image-2 only
• Verified parameter values against OpenAI's official gpt-image-2 docs

Type of Change

• New feature

Testing

Tested manually. Type check passes.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		May 4, 2026 8:26pm

[gitguardian]

⚠️ GitGuardian has uncovered 1 secret following the scan of your pull request.

Please consider investigating the findings and remediating the incidents. Failure to do so may lead to compromising the associated services or software components.

🔎 Detected hardcoded secret in your pull request

GitGuardian id	GitGuardian status	Secret	Commit	Filename
29606901	Triggered	Generic High Entropy Secret	a54dcbe	apps/sim/providers/utils.test.ts	View secret

🛠 Guidelines to remediate hardcoded secrets

1. Understand the implications of revoking this secret by investigating where it is used in your code.
2. Replace and store your secret safely. Learn here the best practices.
3. Revoke and rotate this secret.
4. If possible, rewrite git history. Rewriting git history is not a trivial act. You might completely break other contributing developers' workflow and you risk accidentally deleting legitimate data.

To avoid such incidents in the future consider

• following these best practices for managing and storing secrets including API keys and other credentials
• install secret detection on pre-commit to catch secret before it leaves your machine and ease remediation.

---

^{🦉 GitGuardian detects secrets in your source code to help developers and security teams secure the modern development process. You are seeing this because you or someone else with access to this repository has authorized GitGuardian to scan your pull request.}

[cursor]

PR Summary

Medium Risk
Adds a new OpenAI image model and expands request parameter handling, which could affect image generation behavior and API compatibility if defaults/parameter mappings are wrong. Changes are localized to the image generator block and openai_image tool.

Overview
Adds gpt-image-2 support to the Image Generator block and the openai_image tool, including model selection and updated default size behavior (dall-e-3 defaults to 1024x1024, GPT models default to auto).

Introduces GPT-image-specific options for quality (auto/low/medium/high), outputFormat (png/jpeg/webp), and moderation (auto/low, gpt-image-2 only), and wires these through to the OpenAI images/generations request body (including output_format mapping).

^{Reviewed by Cursor Bugbot for commit b487934. Configure here.}

[waleedlatif1]

Closing — recreating with clean history (single commit only).

[greptile-apps]

Greptile Summary

This PR adds gpt-image-2 as a third model option in the Image Generator block, wiring up its specific parameters (size options up to 4K, quality auto/low/medium/high, background without transparent, output_format, and moderation) while correctly sharing quality/background/outputFormat sub-blocks with gpt-image-1. It also fixes a pre-existing gap where quality and outputFormat were never forwarded for gpt-image-1.

Confidence Score: 4/5

Safe to merge; only P2 cosmetic/observability findings, no correctness or security issues.

All parameter routing is correct: camelCase→snake_case mapping for output_format, moderation gated to gpt-image-2 only, size defaults updated properly. The two findings are a stale log string and missing outputFormat in response metadata — neither affects functional correctness.

No files require special attention.

Important Files Changed

Filename	Overview
apps/sim/blocks/blocks/image_generator.ts	Adds gpt-image-2 model support with correct size/quality/background/outputFormat/moderation sub-blocks; fixes missing quality pass-through for gpt-image-1; size fallback logic updated correctly.
apps/sim/tools/openai/image.ts	Extends the request body builder to handle gpt-image-2; correctly maps camelCase outputFormat to snake_case output_format; adds moderation gating; the URL-response log still hard-codes "DALL-E 3" — minor cosmetic issue only.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Block: params function] -->|model=dall-e-3| B[quality, style → baseParams]
    A -->|model=gpt-image-1| C[quality, background, outputFormat → baseParams]
    A -->|model=gpt-image-2| D[quality, background, outputFormat, moderation → baseParams]

    B --> E[Tool: body builder]
    C --> E
    D --> E

    E -->|dall-e-3| F[body.quality, body.style]
    E -->|gpt-image-1 or gpt-image-2| G[body.quality, body.background, body.output_format]
    G -->|gpt-image-2 only| H[body.moderation]

    F --> I[POST /v1/images/generations]
    G --> I
    H --> I

    I -->|url| J[Fetch via proxy → base64]
    I -->|b64_json| K[Use directly]
    J --> L[Return image + metadata]
    K --> L

Loading

Comments Outside Diff (2)

1. apps/sim/tools/openai/image.ts, line 127-129 (link)

   The URL-response log message still hard-codes "DALL-E 3" even after the modelName variable is already in scope. When gpt-image-2 ever returns a URL response this log will be misleading.

2. apps/sim/tools/openai/image.ts, line 188-197 (link)

   Missing outputFormat in response metadata

   The transformResponse returns metadata: { model } only. When output_format is jpeg or webp, the base64 payload lacks any MIME-type hint, so downstream consumers rendering the raw base64 (e.g. data:image/...;base64,...) won't know the correct content type. Consider echoing outputFormat back through metadata so callers can construct a correct data URI.

_{Reviews (1): Last reviewed commit: "feat(image-generator): add gpt-image-2 m..." | Re-trigger Greptile}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit b487934. Configure here.}

[cursor]

Duplicated params logic for gpt-image-1 and gpt-image-2

Low Severity

The gpt-image-1 and gpt-image-2 branches in the block's params function share three identical lines (quality, background, outputFormat spreads), differing only by the extra moderation line. Meanwhile, the tool's body function in image.ts already consolidates these two models into a single else if branch with a nested check for moderation. The block params function could follow the same consolidated pattern, reducing the risk of future changes being applied to one branch but not the other.

Additional Locations (1)

• apps/sim/tools/openai/image.ts#L96-L104

 

^{Reviewed by Cursor Bugbot for commit b487934. Configure here.}