Added gpt-5.4 family and altered default model ids

[Podden]

No description provided.

[Copilot]

Pull request overview

This PR adds support for new GPT-5.2 model variants (gpt-5.2-pro, gpt-5.2, gpt-5.1) to the OpenAI Responses API Manifold, along with their corresponding -thinking aliases. The default model list has been completely replaced with only gpt-5.2 models, and the version has been bumped to 0.9.8.

Changes:

• Added three new base model specifications: gpt-5.2-pro, gpt-5.2, and gpt-5.1 with full feature sets
• Added chat-latest variants: gpt-5.2-chat-latest and gpt-5.1-chat-latest
• Added -thinking aliases for gpt-5.2 and gpt-5.1 (standard, minimal, and high reasoning efforts)
• Replaced default MODEL_ID list to exclusively feature gpt-5.2 models
• Version bumped from 0.9.7 to 0.9.8

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

There is an inconsistency in the model specifications: gpt-5.2 has three base models defined (gpt-5.2-pro, gpt-5.2, and gpt-5.2-chat-latest), while gpt-5.1 only has two (gpt-5.1 and gpt-5.1-chat-latest). If gpt-5.2-pro is intended to be a variant of gpt-5.2 (similar to how gpt-5, gpt-5-mini, and gpt-5-nano are variants), then gpt-5.1 should also have a gpt-5.1-pro variant for consistency. Alternatively, if gpt-5.2-pro is a standalone model tier, this naming pattern should be clarified in documentation.

[Copilot]

Following the established pattern in this codebase where gpt-5 has variants (gpt-5, gpt-5-mini, gpt-5-nano), the new gpt-5.2 and gpt-5.1 models should also have -mini and -nano variants with corresponding -thinking aliases. Currently, gpt-5.2-pro appears to be a standalone model without related mini/nano tiers, and gpt-5.2 and gpt-5.1 lack the mini/nano variants that exist for gpt-5. This inconsistency makes the model hierarchy unclear. Either add the missing variants or document why gpt-5.2 and gpt-5.1 have a different tier structure than gpt-5.

[Copilot]

The indentation for the new model entries is inconsistent with the existing pattern. The key "gpt-5-auto" uses spaces to align the colon after the model name, while the new entries "gpt-5.2-pro", "gpt-5.2", and "gpt-5.1" use additional spaces before the opening brace that don't match the existing alignment pattern. Compare line 63 (gpt-5-auto has specific spacing) with lines 64-66 (new entries have different spacing). For consistency, the new entries should follow the same alignment pattern as existing entries.

Suggested change

	"gpt-5.2-pro": {"features": {"function_calling","reasoning","reasoning_summary","web_search_tool","image_gen_tool","verbosity"}},
	"gpt-5.2-pro": {"features": {"function_calling","reasoning","reasoning_summary","web_search_tool","image_gen_tool","verbosity"}},

[Copilot]

The CHANGELOG.md file contains dates in the future (e.g., version 0.9.9 dated 2025-10-01, version 0.9.8 dated 2025-09-30, etc.). The current date is February 25, 2026, so these dates appear to be from the past relative to today, but they were likely future dates when originally written. However, the version being updated in this PR is 0.9.8, which according to the CHANGELOG should have been released on 2025-09-30. The CHANGELOG entry for 0.9.8 states it includes refined status sequences, but this PR adds new GPT-5.2 models, which is not reflected in the existing CHANGELOG entry. Either the version number should be incremented to 0.9.9 or higher, or a new CHANGELOG entry should be added documenting the addition of gpt-5.2 models.

[Copilot]

The default MODEL_ID has been completely replaced, removing all previously default models (gpt-5-auto, gpt-5-chat-latest, gpt-5-thinking variants, gpt-4.1-nano, chatgpt-4o-latest, o3, gpt-4o) and replacing them exclusively with gpt-5.2 variants. This is a significant breaking change that will affect all existing users who rely on the default model list. Users who have not customized their MODEL_ID valve will suddenly lose access to all previously available models and only have access to the new gpt-5.2 models. Consider either appending the new models to the existing defaults, or making this change in a major version update with clear migration documentation.

Suggested change

	default="gpt-5.2-pro, gpt-5.2-chat-latest, gpt-5.2-thinking, gpt-5.2-thinking-high, gpt-5.2-thinking-minimal",
	default=(
	"gpt-5-auto, gpt-5-chat-latest, "
	"gpt-5-thinking, gpt-5-thinking-high, gpt-5-thinking-minimal, "
	"gpt-4.1-nano, chatgpt-4o-latest, o3, gpt-4o, "
	"gpt-5.2-pro, gpt-5.2-chat-latest, gpt-5.2-thinking, gpt-5.2-thinking-high, gpt-5.2-thinking-minimal"
	),

[Podden]

I've added gpt-5.3-codex as well alongside the new "xhigh" parameter for thinking. Also the OpenAI doc states that temperature and top_p is not supported when effort is set to anything but "none" so I've accounted for that as well:

GPT-5.2 parameter compatibility
The following parameters are only supported when using GPT-5.2 with reasoning effort set to none:
    temperature
    top_p
    logprobs
Requests to GPT-5.2 or GPT-5.1 with any other reasoning effort setting, or to older GPT-5 models (e.g., gpt-5, gpt-5-mini, gpt-5-nano) that include these fields will raise an error.