[BOT ISSUE] xAI: add missing canonical grok-4.20-multi-agent-0309 model ID

[github-actions]

Gap

The canonical model ID grok-4.20-multi-agent-0309 is missing from model_list.json. Only the alias grok-4.20-multi-agent-beta-0309 is present (line 11601). The xAI docs list grok-4.20-multi-agent-0309 as the primary model and grok-4.20-multi-agent-beta-0309 as an alias of it.

Official sources

Field	Source	URL
Model ID, pricing, context	xAI models listing page	https://docs.x.ai/developers/models
Model ID, alias, pricing	xAI model detail page	https://docs.x.ai/developers/models/grok-4.20-multi-agent-0309

Recommended catalog entry

"grok-4.20-multi-agent-0309": {
  "format": "openai",
  "flavor": "chat",
  "multimodal": true,
  "input_cost_per_mil_tokens": 1.25,
  "output_cost_per_mil_tokens": 2.5,
  "input_cache_read_cost_per_mil_tokens": 0.2,
  "displayName": "Grok 4.20 Multi-Agent (0309)",
  "reasoning": true,
  "max_input_tokens": 1000000,
  "available_providers": ["xAI"]
}

Notes on specific fields

• max_output_tokens: Intentionally omitted. The existing alias entry has 2000000 but xAI docs do not specify a max output token limit for this model. The alias entry's value is unverified.
• fallback_models: Consider adding ["grok-4.20-multi-agent-beta-0309"] to link the canonical ID to the alias. Conversely, the alias entry should get parent: "grok-4.20-multi-agent-0309" to indicate it points to this canonical model.

Verification checklist

• Cross-source: Model ID and pricing confirmed on two independent official xAI pages (models listing + model detail page)
• Already-fixed check: Grepped model_list.json for grok-4.20-multi-agent-0309 — no matches. Only the alias grok-4.20-multi-agent-beta-0309 exists.
• ID format check: grok-4.20-multi-agent-0309 matches the xAI naming convention used by all other grok-4.20 variants in the catalog (e.g., grok-4.20-0309-reasoning, grok-4.20-0309-non-reasoning)
• Duplicate issue check: Searched open issues for "grok-4.20-multi-agent" — only [BOT ISSUE] xAI: add missing input_cache_read_cost_per_mil_tokens to grok-4.20-0309-non-reasoning #978 exists, which covers a different gap (missing cached pricing on grok-4.20-0309-non-reasoning)
• Not deprecated: Model is active with no announced deprecation date
• Not embedding/CV-only: This is a chat model

Comparison with existing alias entry

Field	Alias (beta-0309)	Canonical (0309) proposed
format	openai	openai
flavor	chat	chat
multimodal	true	true
input_cost_per_mil_tokens	1.25	1.25
output_cost_per_mil_tokens	2.5	2.5
input_cache_read_cost_per_mil_tokens	0.2	0.2
reasoning	true	true
max_input_tokens	1000000	1000000
max_output_tokens	2000000	(omitted — unverified)
available_providers	["xAI"]	["xAI"]