update gemini embedding model cost metrics

[ekeith (evanmkeith)]

Add cost metrics to Gemini embedding models

Adds input_cost_per_mil_tokens and output_cost_per_mil_tokens to the Gemini embedding models in model_list.json, covering both the Gemini API entries and the Vertex (publishers/google/models/...) entries:

Model	Input ($/1M tokens)	Output ($/1M tokens)
gemini-embedding-001	0.15	0
gemini-embedding-2	0.20	0

Pricing verified against the official Gemini API pricing page. Embedding models have no billed output tokens, so output cost is set to 0.

Note: gemini-embedding-2 is multimodal with per-modality input pricing (text $0.20/1M, image $0.45/1M, audio $6.50/1M, video $12.00/1M). The schema only supports a single input cost, so these entries reflect the text rate; non-text embedding inputs will be under-costed until per-modality cost fields exist.

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
ai-proxy	Ready	Preview, Comment	Jul 22, 2026 4:46pm

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7d40e1f4dd

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "Codex (@codex) address that feedback".

[chatgpt-codex-connector]

Avoid using the text rate for multimodal embeddings

For gemini-embedding-2 this flat $0.20/M rate only matches text inputs, while the same entry is still marked multimodal and Google's pricing page lists higher image/audio/video input rates (https://ai.google.dev/gemini-api/docs/pricing#gemini-embedding-2). Because the schema exposes only one input_cost_per_mil_tokens value and billing/usage records aggregate input tokens, any image, audio, video, or PDF embedding request routed through this model will be costed as text instead of the modality-specific rate, under-reporting cost for those production requests; consider omitting this cost until modality-aware pricing is represented or restricting it to text-only entries.

Useful? React with 👍 / 👎.