fix: add Google models gemini-2.5-computer-use-preview-10-2025

[github-actions]

fix: add Google models gemini-2.5-computer-use-preview-10-2025

Closes #536

Source issue: #536

Summary

Field	Value
Provider	google
Primary model	gemini-2.5-computer-use-preview-10-2025
Changed models	gemini-2.5-computer-use-preview-10-2025
Added models	gemini-2.5-computer-use-preview-10-2025
Updated models	None
Verification sources	1 2 3 4 5 6 7 8

Verified metadata

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
gemini-2.5-computer-use-preview-10-2025	Gemini 2.5 Computer Use Preview (10-2025)		google	google	chat	input=128000, output=64000	in/out=1.25/10 per 1M	multimodal=true

Verification notes

Verification

Official sources consulted

1. https://ai.google.dev/gemini-api/docs/models — Confirmed model ID gemini-2.5-computer-use-preview-10-2025 exists under "Tool and agent models." Described as a specialized model for browser control agents.
2. https://ai.google.dev/gemini-api/docs/computer-use — Confirmed token limits: input token limit 128,000; output token limit 64,000. Confirmed input types: image + text; output: text.
3. https://ai.google.dev/pricing — Confirmed pricing: $1.25 per 1M input tokens (prompts <= 200k), $10.00 per 1M output tokens (prompts <= 200k). No context caching pricing available. Free tier not available.
4. https://cloud.google.com/vertex-ai/generative-ai/pricing — Confirmed same pricing on Vertex AI ($1.25 input, $10.00 output per 1M tokens). Notes that "Computer Use billing uses the Gemini 2.5 Pro SKU."

Fields verified

Field	Value	Source
format	google	Inferred from provider (Google AI)
flavor	chat	Model takes text/image input and produces text output
multimodal	true	ai.google.dev/gemini-api/docs/computer-use — accepts image + text input
input_cost_per_mil_tokens	1.25	ai.google.dev/pricing
output_cost_per_mil_tokens	10	ai.google.dev/pricing
max_input_tokens	128000	ai.google.dev/gemini-api/docs/computer-use
max_output_tokens	64000	ai.google.dev/gemini-api/docs/computer-use
available_providers	["google"]	Confirmed on Google AI; Vertex variant not verified for regions/locations
displayName	Gemini 2.5 Computer Use Preview (10-2025)	Follows catalog convention for dated preview models

Fields omitted (not published or not applicable)

• input_cache_read_cost_per_mil_tokens: Google pricing page explicitly shows cache pricing as "Not available" for this model.
• input_cache_write_cost_per_mil_tokens: Same — not available.
• reasoning / reasoning_budget: Not explicitly documented for this model variant. The pricing page mentions "response and reasoning" in output cost, but the computer-use docs do not confirm reasoning/thinking token support as a configurable feature. Omitted to avoid guessing.
• parent: No stable alias gemini-2.5-computer-use exists in the catalog or in Google's documentation. This is a standalone preview model.
• supported_regions: Not confirmed for the Google AI provider entry specifically.
• deprecation_date: Not published.

sync_models (LiteLLM) cross-check

The model gemini-2.5-computer-use-preview-10-2025 is not present in the LiteLLM model_prices_and_context_window_backup.json catalog. Therefore no numeric field comparison is possible. All proposed values are sourced directly from official Google documentation.

sync_models vs proposed update

sync_models cross-check found no pricing/token discrepancies for the changed models.

No sync_models discrepancies were noted.

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
ai-proxy	Ready	Preview, Comment	May 7, 2026 9:07am

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c72c63261b

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]

Register the Vertex publisher model too

For Vertex credentials, fetchOpenAI only hands Google models to the Gemini path when the requested model starts with publishers/google/models/ (checked packages/proxy/src/proxy.ts), and the existing Vertex Gemini entries are registered under that publisher-prefixed name. Google Cloud's Vertex Computer Use docs list gemini-2.5-computer-use-preview-10-2025, so with only the short google mapping here, Vertex users cannot route this newly added model through the supported path; add the corresponding publishers/google/models/gemini-2.5-computer-use-preview-10-2025 model/list entry and ['vertex'] endpoint mapping.

Useful? React with 👍 / 👎.

[github-actions]

Proxy model verification

Verification failed for one or more models.

Passed: 0
Failed: 1

Model	Status
gemini-2.5-computer-use-preview-10-2025	❌ Failed

View job details