[BOT ISSUE] Bedrock: add missing pricing for openai.gpt-oss-120b-1:0 and openai.gpt-oss-20b-1:0

[github-actions]

Summary

Two Bedrock OpenAI GPT-OSS models were added to the catalog (via issue #697) but are missing input_cost_per_mil_tokens and output_cost_per_mil_tokens. AWS publishes on-demand pricing on the Bedrock pricing page. The safeguard variants of the same models (openai.gpt-oss-safeguard-120b at $0.15/$0.60, openai.gpt-oss-safeguard-20b at $0.07/$0.20) already have correct pricing in the catalog, confirming Bedrock does publish per-token pricing for this model family.

Affected Models

Model ID	Expected Input $/1M	Expected Output $/1M	Line
openai.gpt-oss-120b-1:0	~0.15	~0.60	~12568
openai.gpt-oss-20b-1:0	~0.07	~0.30	~12578

Both entries currently have format: "converse", flavor: "chat", correct displayName, max_input_tokens: 128000, max_output_tokens: 16384, and available_providers: ["bedrock"] — only pricing fields are missing.

Verification

Check	Status	Detail
Cross-source (model existence)	✅	Both confirmed on Bedrock model cards AND Bedrock pricing page AND already present in model_list.json
Cross-source (pricing)	✅	Asia Pacific (Sydney) pricing confirmed via WebFetch: GPT-OSS 120B at $0.1545/$0.6180, GPT-OSS 20B at $0.0721/$0.3090 per 1M tokens. Fireworks variants (accounts/fireworks/models/gpt-oss-120b at $0.15/$0.60, accounts/fireworks/models/gpt-oss-20b at $0.07/$0.30) provide consistent reference
Sibling consistency	✅	Safeguard variants on Bedrock have pricing: openai.gpt-oss-safeguard-120b ($0.15/$0.60), openai.gpt-oss-safeguard-20b ($0.07/$0.20)
Standard tier available	✅	Issue #697 confirmed Standard, Priority, Flex tiers
Duplicate check	✅	No open issue covers Bedrock GPT-OSS base model pricing. Issue #697 (closed) originally added the models without pricing

Verification Notes

Important: The exact US East (N. Virginia) pricing should be verified directly from the Bedrock pricing page. The values listed above are from the Asia Pacific (Sydney) region and Fireworks cross-reference. Sydney pricing is typically slightly above US pricing (cf. deepseek.v3.2 at $0.62/$1.85 US vs $0.5974/$1.7304 Sydney shows ~3-6% regional markup).

Local Files Inspected

• packages/proxy/schema/model_list.json — confirmed both entries lack pricing fields (lines ~12568, ~12578)
• Safeguard variants at lines ~12588, ~12600 have pricing for reference
• packages/proxy/scripts/sync_models.ts — accounts/fireworks/models/gpt-oss-20b is in SYNC_PRESERVED_FIELDS (Fireworks entry); the Bedrock entries are not preserved

{
  "kind": "cost_update",
  "provider": "bedrock",
  "models": ["openai.gpt-oss-120b-1:0", "openai.gpt-oss-20b-1:0"],
  "status": "needs_fix",
  "model_specs": {
    "openai.gpt-oss-120b-1:0": {
      "input_cost_per_mil_tokens": 0.15,
      "output_cost_per_mil_tokens": 0.60
    },
    "openai.gpt-oss-20b-1:0": {
      "input_cost_per_mil_tokens": 0.07,
      "output_cost_per_mil_tokens": 0.30
    }
  },
  "source_urls": [
    "https://aws.amazon.com/bedrock/pricing/",
    "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-openai-gpt-oss-120b.html",
    "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-openai-gpt-oss-20b.html"
  ]
}

[github-actions]

Autofix skipped because the issue body/title did not contain parseable bot metadata for a model catalog change.