[BOT ISSUE] Vertex: add missing Meta Llama 4 Maverick and Scout MaaS entries

[github-actions]

Summary

Google Cloud Vertex AI offers Meta Llama 4 Maverick and Llama 4 Scout as managed MaaS (Model-as-a-Service) partner models with per-token pricing, but the catalog has no publishers/meta/models/llama-4-* entries. The existing publishers/meta/models/llama-3.*-maas entries (lines ~7390–7420) demonstrate the naming convention and entry format; the Llama 4 equivalents are absent.

Note: There are also no base llama-4-* entries in the catalog at all — these models are entirely missing from the catalog across all providers.

Missing Models

Model ID	Display Name	Provider	Pricing ($/1M tokens)
publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas	Llama 4 Maverick (17B×128E)	vertex	$0.50 input / $1.50 output
publishers/meta/models/llama-4-scout-17b-16e-instruct-maas	Llama 4 Scout (17B×16E)	vertex	$0.20 input / $0.60 output

Suggested Catalog Entries

"publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas": {
  "format": "openai",
  "flavor": "chat",
  "multimodal": true,
  "input_cost_per_mil_tokens": 0.5,
  "output_cost_per_mil_tokens": 1.5,
  "displayName": "Llama 4 Maverick (17Bx128E) Instruct",
  "available_providers": [
    "vertex"
  ]
}

"publishers/meta/models/llama-4-scout-17b-16e-instruct-maas": {
  "format": "openai",
  "flavor": "chat",
  "multimodal": true,
  "input_cost_per_mil_tokens": 0.2,
  "output_cost_per_mil_tokens": 0.6,
  "displayName": "Llama 4 Scout (17Bx16E) Instruct",
  "available_providers": [
    "vertex"
  ]
}

Token limits omitted — the Google Cloud model detail pages could not be scraped to confirm exact context window and max output values. The Llama 4 Maverick is documented at 1M tokens context and Llama 4 Scout at 10M tokens context on Meta's model card, but these should be verified against Vertex's actual limits before setting.

Sources

• Google Cloud Vertex AI Partner Models — lists Llama 4 Maverick and Llama 4 Scout under Partner Models > Meta Llama
• Vertex AI Pricing — lists meta/llama-4-maverick-17b-128e-instruct-maas at $0.50/$1.50 and meta/llama-4-scout-17b-16e-instruct-maas at $0.20/$0.60

Verification Checklist

Check	Status	Detail
Cross-source	✅	Partner models listing page confirms availability; pricing page confirms exact model IDs and per-token pricing
Already fixed	✅	grep for llama-4 and llama4 in model_list.json returns zero matches — no Llama 4 entries exist
ID format	✅	Follows existing publishers/meta/models/llama-3.3-70b-instruct-maas convention (line ~7390)
Duplicate check	✅	No open issue covers Llama 4 Vertex entries

Verification Notes

Field	Source	Notes
Model IDs	Vertex AI pricing page	meta/llama-4-maverick-17b-128e-instruct-maas and meta/llama-4-scout-17b-16e-instruct-maas; catalog key adds publishers/ + /models/ prefix per existing convention
format: "openai"	Existing publishers/meta/models/ entries	All use openai format
flavor: "chat"	Existing entries + Llama 4 model cards	Chat/instruction-tuned models
multimodal: true	Meta Llama 4 announcement	Both Maverick and Scout support image input
Pricing	Vertex AI pricing page	Maverick: $0.50/$1.50, Scout: $0.20/$0.60 per 1M tokens
Token limits	Not confirmed from Vertex docs	Meta's model cards list 1M (Maverick) and 10M (Scout) context, but Vertex may impose different limits

Local Files Inspected

• packages/proxy/schema/model_list.json — grep for llama-4 returns 0 matches; existing publishers/meta/models/llama-3.* entries at lines ~7390–7420

{
  "kind": "missing_model",
  "provider": "vertex",
  "models": ["publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas", "publishers/meta/models/llama-4-scout-17b-16e-instruct-maas"],
  "status": "active",
  "model_specs": {
    "publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "input_cost_per_mil_tokens": 0.5,
      "output_cost_per_mil_tokens": 1.5,
      "displayName": "Llama 4 Maverick (17Bx128E) Instruct",
      "available_providers": ["vertex"]
    },
    "publishers/meta/models/llama-4-scout-17b-16e-instruct-maas": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "input_cost_per_mil_tokens": 0.2,
      "output_cost_per_mil_tokens": 0.6,
      "displayName": "Llama 4 Scout (17Bx16E) Instruct",
      "available_providers": ["vertex"]
    }
  },
  "source_urls": [
    "https://docs.cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-partner-models",
    "https://cloud.google.com/vertex-ai/generative-ai/pricing"
  ]
}

Summary: Two Meta Llama 4 MaaS models are available on Vertex AI with confirmed pricing but have no entries in the catalog. No Llama 4 entries exist in the catalog under any provider.

[github-actions]

Prepared catalog updates for publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas, publishers/meta/models/llama-4-scout-17b-16e-instruct-maas. Added models: publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas, publishers/meta/models/llama-4-scout-17b-16e-instruct-maas.

Verification sources: https://docs.cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-partner-models, https://cloud.google.com/vertex-ai/generative-ai/pricing, https://developers.googleblog.com/en/llama-4-ga-maas-vertex-ai/, https://huggingface.co/meta-llama/Llama-4-Maverick-17B-128E-Instruct, https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct
Verified metadata applied:

• publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas: format=openai, flavor=chat, displayName=Llama 4 Maverick (17Bx128E) Instruct, max_input_tokens=1048576, max_output_tokens=not provided, pricing(input/output per 1M)=0.5/1.5, providers=vertex, locations=us-east5, multimodal=true
• publishers/meta/models/llama-4-scout-17b-16e-instruct-maas: format=openai, flavor=chat, displayName=Llama 4 Scout (17Bx16E) Instruct, max_input_tokens=10485760, max_output_tokens=not provided, pricing(input/output per 1M)=0.2/0.6, providers=vertex, locations=us-east5, multimodal=true

Included in the daily batch PR: #989

[github-actions]

Prepared catalog updates for publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas, publishers/meta/models/llama-4-scout-17b-16e-instruct-maas. Added models: publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas, publishers/meta/models/llama-4-scout-17b-16e-instruct-maas.

Verification sources: https://docs.cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-partner-models, https://cloud.google.com/vertex-ai/generative-ai/pricing, https://docs.cloud.google.com/vertex-ai/generative-ai/docs/partner-models/llama/llama4-maverick, https://docs.cloud.google.com/vertex-ai/generative-ai/docs/partner-models/llama/llama4-scout
Verified metadata applied:

• publishers/meta/models/llama-4-maverick-17b-128e-instruct-maas: format=openai, flavor=chat, displayName=Llama 4 Maverick (17Bx128E) Instruct, max_input_tokens=524288, max_output_tokens=8192, pricing(input/output per 1M)=0.5/1.5, providers=vertex, locations=us-east5, supported_regions=us-east5, multimodal=true
• publishers/meta/models/llama-4-scout-17b-16e-instruct-maas: format=openai, flavor=chat, displayName=Llama 4 Scout (17Bx16E) Instruct, max_input_tokens=1310720, max_output_tokens=8192, pricing(input/output per 1M)=0.2/0.6, providers=vertex, locations=us-east5, supported_regions=us-east5, multimodal=true

Included in the daily batch PR: #1026