chore(cognitive): update AI model catalog

[github-actions]

Model Update Summary

Updated 20 models across 6 providers (anthropic, google-ai, groq, cerebras, xai, fireworks-ai). OpenRouter had no changes.

---

Anthropic

defaultModel updated: claude-sonnet-4-5-20250929 → claude-sonnet-4-6

New models added

Model ID	Display Name	Context	Max Output	Input $/1M	Output $/1M	Lifecycle
claude-opus-4-7	Claude Opus 4.7	1,000,000	128,000	$5.00	$25.00	production
claude-sonnet-4-6	Claude Sonnet 4.6	1,000,000	64,000	$3.00	$15.00	production
claude-opus-4-6	Claude Opus 4.6	1,000,000	128,000	$5.00	$25.00	production
claude-opus-4-5-20251101	Claude Opus 4.5	200,000	64,000	$5.00	$25.00	production
claude-opus-4-1-20250805	Claude Opus 4.1	200,000	32,000	$15.00	$75.00	production
claude-opus-4-20250514	Claude Opus 4	200,000	32,000	$15.00	$75.00	deprecated

Updated models

Model ID	What changed	Old value → New value
claude-sonnet-4-20250514	lifecycle	production → deprecated
claude-sonnet-4-20250514	deprecationDate added	— → 2026-06-15
claude-sonnet-4-20250514	replacementModels added	— → ['claude-sonnet-4-6']
claude-sonnet-4-20250514	tags	included recommended → replaced with deprecated

---

Google AI

New models added

Model ID	Display Name	Context	Max Output	Input $/1M	Output $/1M	Lifecycle
gemini-3.1-pro	Gemini 3.1 Pro	1,048,576	65,536	$2.00	$12.00	preview
gemini-2.5-flash-lite	Gemini 2.5 Flash Lite	1,048,576	65,536	$0.10	$0.40	production

Updated models

Model ID	What changed	Old value → New value
gemini-3-pro	lifecycle	preview → discontinued
gemini-3-pro	discontinuedDate added	— → 2026-03-09
gemini-3-pro	replacementModels added	— → ['gemini-3.1-pro']
gemini-2.0-flash	lifecycle	production → deprecated
gemini-2.0-flash	deprecationDate added	— → 2026-06-01
gemini-2.0-flash	replacementModels added	— → ['gemini-2.5-flash']

---

Groq

New models added

Model ID	Display Name	Context	Max Output	Input $/1M	Output $/1M	Lifecycle
llama-4-scout-17b-16e-instruct	Llama 4 Scout 17B (Preview)	131,072	8,192	$0.11	$0.34	preview
qwen3-32b	Qwen3 32B (Preview)	131,072	40,960	$0.29	$0.59	preview

Updated models

Model ID	What changed	Old value → New value
gpt-oss-20b	inputCostPer1mTokens	$0.10 → $0.075
gpt-oss-20b	outputCostPer1mTokens	$0.50 → $0.30
gpt-oss-20b	maxOutputTokens	32,000 → 65,536
gpt-oss-120b	outputCostPer1mTokens	$0.75 → $0.60
gpt-oss-120b	maxOutputTokens	32,000 → 65,536

---

Cerebras

Updated models

Model ID	What changed	Old value → New value
llama3.1-8b	deprecationDate added	— → 2026-05-27

Note: lifecycle remains production until the deprecation date (May 27, 2026). No new preview models were added because Cerebras does not publish per-token pricing for their preview models (qwen-3-235b-a22b-instruct-2507, zai-glm-4.7).

---

xAI

New models added

Model ID	Display Name	Context	Max Output	Input $/1M	Output $/1M	Lifecycle
grok-4.3	Grok 4.3	1,000,000	128,000	$1.25	$2.50	production

---

Fireworks AI

New models added

Model ID	Display Name	Context	Max Output	Input $/1M	Output $/1M	Lifecycle
deepseek-v4-pro	DeepSeek V4 Pro	1,048,576	16,384	$1.74	$3.48	production

---

OpenRouter

No changes. The config intentionally contains only the gpt-oss-120b fallback model; top-model routing is handled by each primary provider config.

---

OpenAI

No changes. The config already contains the full GPT-5 series (5.2, 5.1, 5, 5-mini, 5-nano), o4-mini, o3, gpt-4.1 series, o3-mini, o1, o1-mini, and legacy deprecated models. All pricing was verified to match current published rates.

Note: OpenAI docs returned 403 during this update cycle; pricing verified via OpenRouter model pages.