@zhixuan92/multi-model-agent-core 3.12.7 → 4.0.0

package/dist/model-profiles.json

@@ -4,16 +4,71 @@
         "naming": "claude-{tier}-{major}-{minor} | claude-{major}-{tier}-{date} (3.x legacy)",
         "rateSource": "https://docs.anthropic.com/en/docs/about-claude/models",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "claude", "supportsEffort": true, "inputTokenSoftLimit": 150000, "capabilities": ["web_search", "web_fetch"] },
-        "profiles": [
-            { "prefix": "claude", "tier": "standard", "cost": "medium", "input": 3, "output": 15, "cachedCreation": 3.75, "bestFor": "general Claude tasks (catch-all, Sonnet rates)" },
-            { "prefix": "claude-opus", "tier": "reasoning", "cost": "high", "input": 5, "output": 25, "cachedCreation": 6.25, "bestFor": "high-ambiguity tasks where error cost is high" },
-            { "prefix": "claude-sonnet", "bestFor": "professional coding and tool use" },
-            { "prefix": "claude-haiku", "cost": "low", "input": 1, "output": 5, "cachedCreation": 1.25, "bestFor": "fast low-cost coding and batch work" },
-            { "prefix": "claude-opus-4-1", "input": 15, "output": 75, "cachedCreation": 18.75, "bestFor": "deprecated, retiring 2026-06-15" },
-            { "prefix": "claude-3", "supportsEffort": false, "capabilities": [], "bestFor": "legacy 3.x (deprecated)" },
-            { "prefix": "claude-3-opus", "tier": "reasoning", "cost": "high", "input": 15, "output": 75, "cachedCreation": 18.75, "bestFor": "legacy Claude 3 Opus (deprecated)" },
-            { "prefix": "claude-3-haiku", "cost": "low", "input": 0.25, "output": 1.25, "cachedCreation": 0.3125, "bestFor": "legacy Claude 3 Haiku (deprecated, retiring 2026-04-19)" }
+        "defaults": {
+            "family": "claude",
+            "supportsEffort": true,
+            "inputTokenSoftLimit": 150000
+        },
+        "profiles": [
+            {
+                "prefix": "claude",
+                "tier": "standard",
+                "cost": "medium",
+                "input": 3,
+                "output": 15,
+                "cachedNonRead": 3.75,
+                "bestFor": "general Claude tasks (catch-all, Sonnet rates)"
+            },
+            {
+                "prefix": "claude-opus",
+                "tier": "reasoning",
+                "cost": "high",
+                "input": 5,
+                "output": 25,
+                "cachedNonRead": 6.25,
+                "bestFor": "high-ambiguity tasks where error cost is high"
+            },
+            {
+                "prefix": "claude-sonnet",
+                "bestFor": "professional coding and tool use"
+            },
+            {
+                "prefix": "claude-haiku",
+                "cost": "low",
+                "input": 1,
+                "output": 5,
+                "cachedNonRead": 1.25,
+                "bestFor": "fast low-cost coding and batch work"
+            },
+            {
+                "prefix": "claude-opus-4-1",
+                "input": 15,
+                "output": 75,
+                "cachedNonRead": 18.75,
+                "bestFor": "deprecated, retiring 2026-06-15"
+            },
+            {
+                "prefix": "claude-3",
+                "supportsEffort": false,
+                "bestFor": "legacy 3.x (deprecated)"
+            },
+            {
+                "prefix": "claude-3-opus",
+                "tier": "reasoning",
+                "cost": "high",
+                "input": 15,
+                "output": 75,
+                "cachedNonRead": 18.75,
+                "bestFor": "legacy Claude 3 Opus (deprecated)"
+            },
+            {
+                "prefix": "claude-3-haiku",
+                "cost": "low",
+                "input": 0.25,
+                "output": 1.25,
+                "cachedNonRead": 0.3125,
+                "bestFor": "legacy Claude 3 Haiku (deprecated, retiring 2026-04-19)"
+            }
         ]
     },
     {
@@ -21,18 +76,78 @@
         "naming": "gpt-{major}.{minor}[-{size}] — sizes: pro, mini, nano (no suffix = base)",
         "rateSource": "https://developers.openai.com/api/docs/pricing",
         "rateLookupDate": "2026-04-25",
-        "defaults": { "family": "openai", "supportsEffort": true, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "gpt", "tier": "standard", "cost": "medium", "input": 2.5, "output": 15, "bestFor": "general GPT family tasks (catch-all)" },
-            { "prefix": "gpt-5", "tier": "reasoning", "bestFor": "reasoning-tier coding and agentic workflows", "inputTokenSoftLimit": 1000000, "capabilities": ["web_search"] },
-            { "prefix": "gpt-5.5", "input": 5, "output": 30, "bestFor": "latest GPT-5 flagship" },
-            { "prefix": "gpt-5.5-pro", "cost": "high", "input": 30, "output": 180, "bestFor": "highest-quality reasoning and complex research" },
-            { "prefix": "gpt-5.4", "bestFor": "prior GPT-5 flagship" },
-            { "prefix": "gpt-5.4-pro", "cost": "high", "input": 30, "output": 180, "bestFor": "previous-gen high-end reasoning" },
-            { "prefix": "gpt-5.4-mini", "tier": "standard", "cost": "low", "input": 0.75, "output": 4.5, "bestFor": "fast cost-effective tasks and extraction" },
-            { "prefix": "gpt-5.4-nano", "tier": "standard", "cost": "low", "input": 0.2, "output": 1.25, "bestFor": "cheapest GPT for high-throughput batch work", "capabilities": [] },
-            { "prefix": "gpt-5.3", "input": 1.75, "output": 14, "bestFor": "reasoning and coding at lower cost than 5.4" },
-            { "prefix": "gpt-4", "tier": "standard", "supportsEffort": false, "bestFor": "legacy GPT-4 family" }
+        "defaults": {
+            "family": "openai",
+            "supportsEffort": true,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "gpt",
+                "tier": "standard",
+                "cost": "medium",
+                "input": 2.5,
+                "output": 15,
+                "bestFor": "general GPT family tasks (catch-all)"
+            },
+            {
+                "prefix": "gpt-5",
+                "tier": "reasoning",
+                "bestFor": "reasoning-tier coding and agentic workflows",
+                "inputTokenSoftLimit": 1000000
+            },
+            {
+                "prefix": "gpt-5.5",
+                "input": 5,
+                "output": 30,
+                "bestFor": "latest GPT-5 flagship"
+            },
+            {
+                "prefix": "gpt-5.5-pro",
+                "cost": "high",
+                "input": 30,
+                "output": 180,
+                "bestFor": "highest-quality reasoning and complex research"
+            },
+            {
+                "prefix": "gpt-5.4",
+                "bestFor": "prior GPT-5 flagship"
+            },
+            {
+                "prefix": "gpt-5.4-pro",
+                "cost": "high",
+                "input": 30,
+                "output": 180,
+                "bestFor": "previous-gen high-end reasoning"
+            },
+            {
+                "prefix": "gpt-5.4-mini",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.75,
+                "output": 4.5,
+                "bestFor": "fast cost-effective tasks and extraction"
+            },
+            {
+                "prefix": "gpt-5.4-nano",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.2,
+                "output": 1.25,
+                "bestFor": "cheapest GPT for high-throughput batch work"
+            },
+            {
+                "prefix": "gpt-5.3",
+                "input": 1.75,
+                "output": 14,
+                "bestFor": "reasoning and coding at lower cost than 5.4"
+            },
+            {
+                "prefix": "gpt-4",
+                "tier": "standard",
+                "supportsEffort": false,
+                "bestFor": "legacy GPT-4 family"
+            }
         ]
     },
     {
@@ -40,12 +155,44 @@
         "naming": "gemini-{major}[.{minor}]-{tier}-{suffix} — tiers: pro ($2/$12), flash ($0.50/$3)",
         "rateSource": "https://cloud.google.com/vertex-ai/generative-ai/pricing",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "gemini", "supportsEffort": false, "inputTokenSoftLimit": 200000, "capabilities": ["web_search"] },
-        "profiles": [
-            { "prefix": "gemini", "tier": "standard", "cost": "low", "input": 0.5, "output": 3, "bestFor": "general Gemini tasks (catch-all, Flash rates)" },
-            { "prefix": "gemini-3.1-pro", "tier": "reasoning", "cost": "medium", "input": 2, "output": 12, "bestFor": "long-context reasoning with large documents" },
-            { "prefix": "gemini-3-pro", "tier": "reasoning", "cost": "medium", "input": 2, "output": 12, "bestFor": "Gemini 3.0 Pro variants" },
-            { "prefix": "gemini-2.5-pro", "tier": "reasoning", "cost": "medium", "input": 2, "output": 12, "bestFor": "reasoning and coding with extended thinking" }
+        "defaults": {
+            "family": "gemini",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 200000
+        },
+        "profiles": [
+            {
+                "prefix": "gemini",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.5,
+                "output": 3,
+                "bestFor": "general Gemini tasks (catch-all, Flash rates)"
+            },
+            {
+                "prefix": "gemini-3.1-pro",
+                "tier": "reasoning",
+                "cost": "medium",
+                "input": 2,
+                "output": 12,
+                "bestFor": "long-context reasoning with large documents"
+            },
+            {
+                "prefix": "gemini-3-pro",
+                "tier": "reasoning",
+                "cost": "medium",
+                "input": 2,
+                "output": 12,
+                "bestFor": "Gemini 3.0 Pro variants"
+            },
+            {
+                "prefix": "gemini-2.5-pro",
+                "tier": "reasoning",
+                "cost": "medium",
+                "input": 2,
+                "output": 12,
+                "bestFor": "reasoning and coding with extended thinking"
+            }
         ]
     },
     {
@@ -53,11 +200,36 @@
         "naming": "grok-{major}[.{minor}]-{variant} — fast variant is ~15x cheaper than base",
         "rateSource": "https://docs.x.ai/developers/models",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "grok", "supportsEffort": false, "inputTokenSoftLimit": 200000, "capabilities": ["web_search"] },
-        "profiles": [
-            { "prefix": "grok", "tier": "standard", "cost": "low", "input": 0.2, "output": 0.5, "bestFor": "general Grok tasks (catch-all, fast-tier rates)" },
-            { "prefix": "grok-4", "tier": "reasoning", "cost": "medium", "input": 3, "output": 15, "bestFor": "reasoning and analysis tasks" },
-            { "prefix": "grok-4-1-fast", "tier": "standard", "cost": "low", "input": 0.2, "output": 0.5, "bestFor": "fast cost-effective coding and chat" }
+        "defaults": {
+            "family": "grok",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 200000
+        },
+        "profiles": [
+            {
+                "prefix": "grok",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.2,
+                "output": 0.5,
+                "bestFor": "general Grok tasks (catch-all, fast-tier rates)"
+            },
+            {
+                "prefix": "grok-4",
+                "tier": "reasoning",
+                "cost": "medium",
+                "input": 3,
+                "output": 15,
+                "bestFor": "reasoning and analysis tasks"
+            },
+            {
+                "prefix": "grok-4-1-fast",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.2,
+                "output": 0.5,
+                "bestFor": "fast cost-effective coding and chat"
+            }
         ]
     },
     {
@@ -65,11 +237,30 @@
         "naming": "mistral-{tier}-{version} — tiers: large ($0.50/$1.50), small ($0.10/$0.30)",
         "rateSource": "https://mistral.ai/pricing",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "mistral", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "mistral", "tier": "standard", "cost": "low", "input": 0.5, "output": 1.5, "bestFor": "general Mistral tasks (catch-all, Large rates)" },
-            { "prefix": "mistral-large", "bestFor": "cost-effective frontier-class coding" },
-            { "prefix": "mistral-small", "input": 0.1, "output": 0.3, "bestFor": "fast lightweight tasks and classification" }
+        "defaults": {
+            "family": "mistral",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "mistral",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.5,
+                "output": 1.5,
+                "bestFor": "general Mistral tasks (catch-all, Large rates)"
+            },
+            {
+                "prefix": "mistral-large",
+                "bestFor": "cost-effective frontier-class coding"
+            },
+            {
+                "prefix": "mistral-small",
+                "input": 0.1,
+                "output": 0.3,
+                "bestFor": "fast lightweight tasks and classification"
+            }
         ]
     },
     {
@@ -77,11 +268,36 @@
         "naming": "deepseek-{variant}[-{size}] — V4 sizes: flash, pro; legacy variants are flat-priced",
         "rateSource": "https://api-docs.deepseek.com/quick_start/pricing",
         "rateLookupDate": "2026-04-27",
-        "defaults": { "family": "deepseek", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "deepseek", "tier": "standard", "cost": "low", "input": 0.3, "output": 0.5, "bestFor": "cost-effective coding and reasoning" },
-            { "prefix": "deepseek-v4-flash", "input": 0.14, "output": 0.28, "bestFor": "cheapest DeepSeek V4 with 1M context", "inputTokenSoftLimit": 1000000 },
-            { "prefix": "deepseek-v4-pro", "tier": "reasoning", "cost": "medium", "input": 0.435, "output": 0.87, "bestFor": "DeepSeek V4 thinking-mode reasoning with 1M context (limited-time 75% off until 2026-05-05 15:59 UTC; full price input=1.74 output=3.48)", "inputTokenSoftLimit": 1000000 }
+        "defaults": {
+            "family": "deepseek",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "deepseek",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.3,
+                "output": 0.5,
+                "bestFor": "cost-effective coding and reasoning"
+            },
+            {
+                "prefix": "deepseek-v4-flash",
+                "input": 0.14,
+                "output": 0.28,
+                "bestFor": "cheapest DeepSeek V4 with 1M context",
+                "inputTokenSoftLimit": 1000000
+            },
+            {
+                "prefix": "deepseek-v4-pro",
+                "tier": "reasoning",
+                "cost": "medium",
+                "input": 0.435,
+                "output": 0.87,
+                "bestFor": "DeepSeek V4 thinking-mode reasoning with 1M context (limited-time 75% off until 2026-05-05 15:59 UTC; full price input=1.74 output=3.48)",
+                "inputTokenSoftLimit": 1000000
+            }
         ]
     },
     {
@@ -89,11 +305,32 @@
         "naming": "llama-{major}-{variant} | meta-llama/llama-{major}-{variant} (OpenRouter)",
         "rateSource": "https://openrouter.ai/models",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "llama", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "llama", "tier": "standard", "cost": "low", "input": 0.15, "output": 0.6, "bestFor": "general open-weight model tasks" },
-            { "prefix": "llama-4", "bestFor": "cost-effective open-weight coding" },
-            { "prefix": "meta-llama/", "tier": "standard", "cost": "low", "input": 0.15, "output": 0.6, "bestFor": "open-weight coding via OpenRouter" }
+        "defaults": {
+            "family": "llama",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "llama",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.15,
+                "output": 0.6,
+                "bestFor": "general open-weight model tasks"
+            },
+            {
+                "prefix": "llama-4",
+                "bestFor": "cost-effective open-weight coding"
+            },
+            {
+                "prefix": "meta-llama/",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.15,
+                "output": 0.6,
+                "bestFor": "open-weight coding via OpenRouter"
+            }
         ]
     },
     {
@@ -101,9 +338,20 @@
         "naming": "qwen{major}[.{minor}]-{tier} — no separator after 'qwen'",
         "rateSource": "https://openrouter.ai/qwen/qwen3.5-plus-02-15",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "qwen", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "qwen", "tier": "standard", "cost": "low", "input": 0.26, "output": 1.56, "bestFor": "cost-effective coding and multilingual tasks" }
+        "defaults": {
+            "family": "qwen",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "qwen",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.26,
+                "output": 1.56,
+                "bestFor": "cost-effective coding and multilingual tasks"
+            }
         ]
     },
     {
@@ -111,11 +359,31 @@
         "naming": "glm-{major}[.{minor}] — GLM-5 is ~30% more expensive than GLM-4",
         "rateSource": "https://docs.z.ai/guides/overview/pricing",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "glm", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "glm", "tier": "standard", "cost": "medium", "input": 1, "output": 3.2, "bestFor": "general GLM/Zhipu family tasks" },
-            { "prefix": "glm-5", "bestFor": "bilingual Chinese/English coding" },
-            { "prefix": "glm-4", "cost": "low", "input": 0.6, "output": 2.4, "bestFor": "bilingual tasks at lower cost than GLM-5" }
+        "defaults": {
+            "family": "glm",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "glm",
+                "tier": "standard",
+                "cost": "medium",
+                "input": 1,
+                "output": 3.2,
+                "bestFor": "general GLM/Zhipu family tasks"
+            },
+            {
+                "prefix": "glm-5",
+                "bestFor": "bilingual Chinese/English coding"
+            },
+            {
+                "prefix": "glm-4",
+                "cost": "low",
+                "input": 0.6,
+                "output": 2.4,
+                "bestFor": "bilingual tasks at lower cost than GLM-5"
+            }
         ]
     },
     {
@@ -123,9 +391,20 @@
         "naming": "kimi-k{generation}[.{minor}] | kimi-k{generation}-{variant}",
         "rateSource": "https://platform.kimi.ai/docs/pricing/chat",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "kimi", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "kimi", "tier": "standard", "cost": "low", "input": 0.6, "output": 3, "bestFor": "long-context tasks and coding" }
+        "defaults": {
+            "family": "kimi",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "kimi",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.6,
+                "output": 3,
+                "bestFor": "long-context tasks and coding"
+            }
         ]
     },
     {
@@ -133,9 +412,20 @@
         "naming": "command-r[-plus] — plus variant is flagship",
         "rateSource": "https://cohere.com/pricing",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "cohere", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": ["web_search"] },
-        "profiles": [
-            { "prefix": "command-r", "tier": "standard", "cost": "medium", "input": 2.5, "output": 10, "bestFor": "RAG workflows and grounded generation" }
+        "defaults": {
+            "family": "cohere",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "command-r",
+                "tier": "standard",
+                "cost": "medium",
+                "input": 2.5,
+                "output": 10,
+                "bestFor": "RAG workflows and grounded generation"
+            }
         ]
     },
     {
@@ -143,9 +433,21 @@
         "naming": "MiniMax-M{major}.{minor}[-highspeed] — case-sensitive prefix",
         "rateSource": "https://platform.minimax.io/docs/guides/pricing-paygo",
         "rateLookupDate": "2026-04-17",
-        "defaults": { "family": "minimax", "supportsEffort": true, "inputTokenSoftLimit": 200000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "MiniMax-M2", "tier": "standard", "cost": "low", "input": 0.3, "output": 1.2, "bestFor": "well-scoped coding and agent loops where cost matters", "avoidFor": "highest-stakes ambiguous work that needs top-tier judgment" }
+        "defaults": {
+            "family": "minimax",
+            "supportsEffort": true,
+            "inputTokenSoftLimit": 200000
+        },
+        "profiles": [
+            {
+                "prefix": "MiniMax-M2",
+                "tier": "standard",
+                "cost": "low",
+                "input": 0.3,
+                "output": 1.2,
+                "bestFor": "well-scoped coding and agent loops where cost matters",
+                "avoidFor": "highest-stakes ambiguous work that needs top-tier judgment"
+            }
         ]
     },
     {
@@ -153,9 +455,18 @@
         "naming": "phi-{major}[-{variant}]",
         "rateSource": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "phi", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "phi", "tier": "standard", "cost": "low", "bestFor": "efficient open-weight Microsoft models" }
+        "defaults": {
+            "family": "phi",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "phi",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "efficient open-weight Microsoft models"
+            }
         ]
     },
     {
@@ -163,18 +474,36 @@
         "naming": "gemma-{major}[.{minor}] — open-weight Gemma family",
         "rateSource": "https://ai.google.dev/gemma",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "gemma", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "gemma", "tier": "standard", "cost": "low", "bestFor": "lightweight open-weight Google models" }
+        "defaults": {
+            "family": "gemma",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "gemma",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "lightweight open-weight Google models"
+            }
         ]
     },
     {
         "provider": "yi-01",
         "naming": "yi-{tier}[-{version}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "yi", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "yi", "tier": "standard", "cost": "low", "bestFor": "bilingual Chinese/English open-weight models" }
+        "defaults": {
+            "family": "yi",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "yi",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "bilingual Chinese/English open-weight models"
+            }
         ]
     },
     {
@@ -182,9 +511,18 @@
         "naming": "sonar[-{variant}] — Perplexity's online-grounded models",
         "rateSource": "https://docs.perplexity.ai/guides/pricing",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "sonar", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": ["web_search"] },
-        "profiles": [
-            { "prefix": "sonar", "tier": "standard", "cost": "low", "bestFor": "online-grounded search-augmented tasks" }
+        "defaults": {
+            "family": "sonar",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "sonar",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "online-grounded search-augmented tasks"
+            }
         ]
     },
     {
@@ -192,18 +530,36 @@
         "naming": "nova[-{tier}][-{version}] — Amazon Bedrock native models",
         "rateSource": "https://aws.amazon.com/bedrock/nova/pricing/",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "nova", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "nova", "tier": "standard", "cost": "low", "bestFor": "Amazon Bedrock native multimodal models" }
+        "defaults": {
+            "family": "nova",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "nova",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "Amazon Bedrock native multimodal models"
+            }
         ]
     },
     {
         "provider": "ai21",
         "naming": "jamba-{major}[.{minor}] — Mamba-transformer hybrid",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "jamba", "supportsEffort": false, "inputTokenSoftLimit": 256000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "jamba", "tier": "standard", "cost": "low", "bestFor": "long-context Mamba-transformer hybrid" }
+        "defaults": {
+            "family": "jamba",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 256000
+        },
+        "profiles": [
+            {
+                "prefix": "jamba",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "long-context Mamba-transformer hybrid"
+            }
         ]
     },
     {
@@ -211,9 +567,18 @@
         "naming": "granite-{major}[.{minor}]-{variant}",
         "rateSource": "https://www.ibm.com/granite",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "granite", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "granite", "tier": "standard", "cost": "low", "bestFor": "enterprise open-weight IBM models" }
+        "defaults": {
+            "family": "granite",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "granite",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "enterprise open-weight IBM models"
+            }
         ]
     },
     {
@@ -221,9 +586,18 @@
         "naming": "nemotron-{major}[-{variant}]",
         "rateSource": "https://build.nvidia.com/explore/discover",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "nemotron", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "nemotron", "tier": "standard", "cost": "low", "bestFor": "NVIDIA-optimized open-weight models" }
+        "defaults": {
+            "family": "nemotron",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "nemotron",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "NVIDIA-optimized open-weight models"
+            }
         ]
     },
     {
@@ -231,9 +605,18 @@
         "naming": "dbrx[-{variant}]",
         "rateSource": "https://www.databricks.com/dbrx",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "dbrx", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "dbrx", "tier": "standard", "cost": "low", "bestFor": "mixture-of-experts enterprise models" }
+        "defaults": {
+            "family": "dbrx",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "dbrx",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "mixture-of-experts enterprise models"
+            }
         ]
     },
     {
@@ -241,45 +624,90 @@
         "naming": "arctic[-{variant}]",
         "rateSource": "https://www.snowflake.com/en/blog/snowflake-arctic-open-efficient-foundation-language-models-snowflake/",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "arctic", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "arctic", "tier": "standard", "cost": "low", "bestFor": "Dense-MoE hybrid Snowflake models" }
+        "defaults": {
+            "family": "arctic",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "arctic",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "Dense-MoE hybrid Snowflake models"
+            }
         ]
     },
     {
         "provider": "reka",
         "naming": "reka-{tier}[-{version}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "reka", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "reka", "tier": "standard", "cost": "low", "bestFor": "multimodal frontier models" }
+        "defaults": {
+            "family": "reka",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "reka",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "multimodal frontier models"
+            }
         ]
     },
     {
         "provider": "ai2",
         "naming": "olmo-{major}[.{minor}][-{variant}] — fully open (data + code + weights)",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "olmo", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "olmo", "tier": "standard", "cost": "low", "bestFor": "fully open research models from AI2" }
+        "defaults": {
+            "family": "olmo",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "olmo",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "fully open research models from AI2"
+            }
         ]
     },
     {
         "provider": "nous-research",
         "naming": "hermes-{major}[.{minor}][-{variant}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "hermes", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "hermes", "tier": "standard", "cost": "low", "bestFor": "fine-tuned open-weight reasoning models" }
+        "defaults": {
+            "family": "hermes",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "hermes",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "fine-tuned open-weight reasoning models"
+            }
         ]
     },
     {
         "provider": "microsoft-wizardlm",
         "naming": "wizardlm-{major}[.{minor}][-{variant}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "wizardlm", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "wizardlm", "tier": "standard", "cost": "low", "bestFor": "instruction-following open-weight models" }
+        "defaults": {
+            "family": "wizardlm",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "wizardlm",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "instruction-following open-weight models"
+            }
         ]
     },
     {
@@ -287,54 +715,108 @@
         "naming": "starcoder{generation}[-{variant}]",
         "rateSource": "https://huggingface.co/bigcode",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "starcoder", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "starcoder", "tier": "standard", "cost": "low", "bestFor": "code-generation open-weight models" }
+        "defaults": {
+            "family": "starcoder",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "starcoder",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "code-generation open-weight models"
+            }
         ]
     },
     {
         "provider": "cognitive-computations",
         "naming": "dolphin-{major}[.{minor}][-{variant}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "dolphin", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "dolphin", "tier": "standard", "cost": "low", "bestFor": "uncensored open-weight models" }
+        "defaults": {
+            "family": "dolphin",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "dolphin",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "uncensored open-weight models"
+            }
         ]
     },
     {
         "provider": "openchat",
         "naming": "openchat-{major}[.{minor}][-{variant}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "openchat", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "openchat", "tier": "standard", "cost": "low", "bestFor": "open-source chat-tuned models" }
+        "defaults": {
+            "family": "openchat",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "openchat",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "open-source chat-tuned models"
+            }
         ]
     },
     {
         "provider": "lmsys",
         "naming": "vicuna-{major}[b][-{variant}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "vicuna", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "vicuna", "tier": "standard", "cost": "low", "bestFor": "LLaMA-finetuned chat models from LMSYS" }
+        "defaults": {
+            "family": "vicuna",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "vicuna",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "LLaMA-finetuned chat models from LMSYS"
+            }
         ]
     },
     {
         "provider": "shanghai-ai-lab",
         "naming": "internlm{major}[.{minor}][-{variant}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "internlm", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "internlm", "tier": "standard", "cost": "low", "bestFor": "multilingual Chinese/English open-weight models" }
+        "defaults": {
+            "family": "internlm",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "internlm",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "multilingual Chinese/English open-weight models"
+            }
         ]
     },
     {
         "provider": "baichuan",
         "naming": "baichuan{generation}[-{variant}]",
         "rateLookupDate": "2026-04-29",
-        "defaults": { "family": "baichuan", "supportsEffort": false, "inputTokenSoftLimit": 128000, "capabilities": [] },
-        "profiles": [
-            { "prefix": "baichuan", "tier": "standard", "cost": "low", "bestFor": "Chinese-language open-weight models" }
+        "defaults": {
+            "family": "baichuan",
+            "supportsEffort": false,
+            "inputTokenSoftLimit": 128000
+        },
+        "profiles": [
+            {
+                "prefix": "baichuan",
+                "tier": "standard",
+                "cost": "low",
+                "bestFor": "Chinese-language open-weight models"
+            }
         ]
     }
 ]