@newrelic/preflight 1.67.3 → 1.69.0

package/dist/shared/pricing-data.js

@@ -15,6 +15,7 @@ export const MODEL_ALIASES = {
     // Anthropic Claude families
     'claude-opus-4': 'claude-opus-4-8',
     'claude-sonnet-4': 'claude-sonnet-4-6',
+    // Deliberately stays on 4.5 even though Haiku 5.5 exists.
     'claude-haiku-4': 'claude-haiku-4-5',
     'claude-haiku-5': 'claude-haiku-5-5',
     'claude-haiku-3-5': 'claude-haiku-3-5-20241022',
@@ -197,7 +198,9 @@ export const MODEL_ALIASES = {
 // Built-in pricing table — USD per million tokens
 //
 // Rates last fully verified against vendor public pricing pages on
-// 2026-08-27. Later partial passes: Anthropic 5.5, GPT-6, Bedrock, xAI,
+// 2026-10-26 (every vendor group, incl. xAI, Moonshot and MAI-Code; Copilot
+// read by search only). Earlier full
+// pass 2026-08-27. Partial passes since then: Anthropic 5.5, GPT-6, Bedrock, xAI,
 // Moonshot, Mistral Large 4 and the Copilot stale-entry notes on 2026-10-07;
 // OpenAI long-context cache tiers, xAI cache tiers and Moonshot cache writes
 // on 2026-10-08:
@@ -239,6 +242,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 50,
         cacheReadPerMTok: 1,
         cacheCreationPerMTok: 12.5,
+        cacheCreation1hPerMTok: 20,
         contextWindow: 1_000_000,
     },
     // Shares Fable 5's specs and pricing exactly, but is limited-availability
@@ -250,6 +254,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 50,
         cacheReadPerMTok: 1,
         cacheCreationPerMTok: 12.5,
+        cacheCreation1hPerMTok: 20,
         contextWindow: 1_000_000,
     },
     // Extends Fable 5 at the same input/output prices, with cache reads at a
@@ -262,6 +267,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 50,
         cacheReadPerMTok: 0.25,
         cacheCreationPerMTok: 12.5,
+        cacheCreation1hPerMTok: 20,
         contextWindow: 1_000_000,
     },
     // Shares Fable 5.1's specs and pricing exactly (same discounted 0.025x
@@ -273,6 +279,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 50,
         cacheReadPerMTok: 0.25,
         cacheCreationPerMTok: 12.5,
+        cacheCreation1hPerMTok: 20,
         contextWindow: 1_000_000,
     },
     // Opus 5.5 is the newest Opus: $4/$20 (below Opus 5's $5/$25), with cache
@@ -284,6 +291,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 20,
         cacheReadPerMTok: 0.2,
         cacheCreationPerMTok: 5,
+        cacheCreation1hPerMTok: 8,
         contextWindow: 1_000_000,
     },
     'claude-opus-5': {
@@ -292,6 +300,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10,
         contextWindow: 1_000_000,
     },
     'claude-opus-4-8': {
@@ -300,6 +309,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10,
         contextWindow: 1_000_000,
     },
     // $2/$10 is Anthropic's standing price, not introductory — the "What's new
@@ -312,15 +322,18 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 10,
         cacheReadPerMTok: 0.2,
         cacheCreationPerMTok: 2.5,
+        cacheCreation1hPerMTok: 4,
         contextWindow: 1_000_000,
     },
-    // Same rates as claude-sonnet-5.
+    // Same rates as claude-sonnet-5 except cache reads, which are 0.05x input
+    // (vs 0.1x) per Anthropic's pricing page.
     'claude-sonnet-5-5': {
         inputPerMTok: 2,
         outputPerMTok: 10,
         thinkingPerMTok: 10,
-        cacheReadPerMTok: 0.2,
+        cacheReadPerMTok: 0.1,
         cacheCreationPerMTok: 2.5,
+        cacheCreation1hPerMTok: 4,
         contextWindow: 1_000_000,
     },
     'claude-sonnet-4-6': {
@@ -329,25 +342,28 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 15,
         cacheReadPerMTok: 0.3,
         cacheCreationPerMTok: 3.75,
+        cacheCreation1hPerMTok: 6,
         contextWindow: 1_000_000,
     },
     // Haiku 5.5 is priced by prompt length, flat: a prompt over 100k tokens bills
     // the whole request at 5x input/output and 5x cache reads (platform.claude.com
-    // pricing, verified 2026-10-08). The 5m cache-write tier rate ($0.625 above
-    // 100k) is not modeled; cache writes stay at the base $0.125. 1M context per
-    // the model overview. 'claude-haiku-4' deliberately stays on 4.5.
+    // pricing, verified 2026-10-08). The 5m cache-write rate is $0.125, or $0.625
+    // above 100k (tierCacheCreationPerMTok). 1M context per the model overview.
     'claude-haiku-5-5': {
         inputPerMTok: 0.1,
         outputPerMTok: 0.5,
         thinkingPerMTok: 0.5,
         cacheReadPerMTok: 0.01,
         cacheCreationPerMTok: 0.125,
+        cacheCreation1hPerMTok: 0.2,
         contextWindow: 1_000_000,
         tierThreshold: 100_000,
         tierInputPerMTok: 0.5,
         tierOutputPerMTok: 2.5,
         tierThinkingPerMTok: 2.5,
         tierCacheReadPerMTok: 0.05,
+        tierCacheCreationPerMTok: 0.625,
+        tierCacheCreation1hPerMTok: 1,
     },
     // Dateless current-gen entry — MODEL_ALIASES['claude-haiku-4'] routes here.
     // Matches the opus/sonnet convention: alias → dateless key → dated legacy key.
@@ -359,6 +375,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 5,
         cacheReadPerMTok: 0.1,
         cacheCreationPerMTok: 1.25,
+        cacheCreation1hPerMTok: 2,
         contextWindow: 200_000,
     },
     // Anthropic lists retirement "not sooner than" 2026-10-15; recheck the
@@ -369,6 +386,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 5,
         cacheReadPerMTok: 0.1,
         cacheCreationPerMTok: 1.25,
+        cacheCreation1hPerMTok: 2,
         contextWindow: 200_000,
     },
     // ---- Anthropic (legacy Claude 4 generation) ----
@@ -384,6 +402,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10,
         contextWindow: 1_000_000,
     },
     'claude-opus-4-6': {
@@ -392,6 +411,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10,
         contextWindow: 1_000_000,
     },
     // Deprecated 2026-09-30, retires 2026-11-30 (replacement claude-sonnet-5-5).
@@ -402,6 +422,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 15,
         cacheReadPerMTok: 0.3,
         cacheCreationPerMTok: 3.75,
+        cacheCreation1hPerMTok: 6,
         contextWindow: 200_000,
     },
     'claude-opus-4-5': {
@@ -410,6 +431,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10,
         contextWindow: 200_000,
     },
     // This key uses a version-number suffix (-1), not an 8-digit date suffix,
@@ -425,6 +447,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 75,
         cacheReadPerMTok: 1.5,
         cacheCreationPerMTok: 18.75,
+        cacheCreation1hPerMTok: 30,
         contextWindow: 200_000,
     },
     // Retired 2026-06-15. Kept for historical-cost backfill.
@@ -434,6 +457,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 15,
         cacheReadPerMTok: 0.3,
         cacheCreationPerMTok: 3.75,
+        cacheCreation1hPerMTok: 6,
         contextWindow: 200_000,
     },
     // Retired 2026-06-15. Kept for historical-cost backfill.
@@ -443,6 +467,7 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 75,
         cacheReadPerMTok: 1.5,
         cacheCreationPerMTok: 18.75,
+        cacheCreation1hPerMTok: 30,
         contextWindow: 200_000,
     },
     // Retired 2026-02-19. Kept for historical-cost backfill.
@@ -451,6 +476,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 4,
         cacheReadPerMTok: 0.08,
         cacheCreationPerMTok: 1,
+        cacheCreation1hPerMTok: 1.6,
         contextWindow: 200_000,
     },
     // ---- Google Gemini (current generation) ----
@@ -482,6 +508,19 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 3.75,
         contextWindow: 1_000_000,
     },
+    // Gemini 4 Argon: flagship reasoning model. Introductory $2/$10 (cache read
+    // $0.10) through 2026-11-30; Google lists $4/$20 (cache $0.20) from
+    // 2026-12-01, so bump these rates once that date passes. Output price includes
+    // thinking tokens. Explicit caching is unsupported (implicit only), and Google
+    // publishes no context window (1M is its stated long-decode ceiling), so
+    // contextWindow is best-effort. Verified 2026-10-26.
+    'gemini-4-argon': {
+        inputPerMTok: 2,
+        cacheReadPerMTok: 0.1,
+        outputPerMTok: 10,
+        thinkingPerMTok: 10,
+        contextWindow: 1_000_000,
+    },
     'gemini-3.6-flash': {
         inputPerMTok: 0.75,
         cacheReadPerMTok: 0.075,
@@ -489,6 +528,9 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 3.75,
         contextWindow: 1_000_000,
     },
+    // No longer on Google's pricing page: the endpoint auto-routes to
+    // gemini-3.6-flash (deprecations page, 2026-10-26). Google doesn't say which
+    // rate that routing bills, so this entry keeps the historical 3.5 rates.
     'gemini-3.5-flash': {
         inputPerMTok: 1.5,
         cacheReadPerMTok: 0.15,
@@ -527,6 +569,52 @@ export const DEFAULT_PRICING_TABLE = {
     // thinking is NOT a subset of it (extractGeminiTokens() reports
     // thoughtsTokenCount as a separate addend), so thinkingPerMTok must be
     // set explicitly or reasoning spend is silently billed at $0.
+    // Live / native-audio models bill text and audio separately. Input: text
+    // $0.75, audio $3.00 (image/video $1.00 is not modelled and bills at the text
+    // rate). Output $4.50 text / $12.00 audio, thinking at the text rate. No
+    // caching. 131,072 is the published input token limit
+    // (ai.google.dev/gemini-api/docs/pricing, verified 2026-10-26).
+    'gemini-3.8-live': {
+        inputPerMTok: 0.75,
+        audioInputPerMTok: 3,
+        outputPerMTok: 4.5,
+        audioOutputPerMTok: 12,
+        thinkingPerMTok: 4.5,
+        contextWindow: 131_072,
+    },
+    'gemini-3.8-live-extended-thinking': {
+        inputPerMTok: 0.75,
+        audioInputPerMTok: 3,
+        outputPerMTok: 4.5,
+        audioOutputPerMTok: 12,
+        thinkingPerMTok: 4.5,
+        contextWindow: 131_072,
+    },
+    'gemini-3.1-flash-live-preview': {
+        inputPerMTok: 0.75,
+        audioInputPerMTok: 3,
+        outputPerMTok: 4.5,
+        audioOutputPerMTok: 12,
+        thinkingPerMTok: 4.5,
+        contextWindow: 131_072,
+    },
+    'gemini-2.5-flash-native-audio-preview-12-2025': {
+        inputPerMTok: 0.5,
+        audioInputPerMTok: 3,
+        outputPerMTok: 2,
+        audioOutputPerMTok: 12,
+        thinkingPerMTok: 2,
+        contextWindow: 131_072,
+    },
+    // Audio in, audio out at one rate each ($3.50 / $21.00).
+    'gemini-3.5-live-translate-preview': {
+        inputPerMTok: 3.5,
+        audioInputPerMTok: 3.5,
+        outputPerMTok: 21,
+        audioOutputPerMTok: 21,
+        thinkingPerMTok: 21,
+        contextWindow: 131_072,
+    },
     'gemini-3.1-flash-lite': {
         inputPerMTok: 0.25,
         audioInputPerMTok: 0.5,
@@ -558,7 +646,17 @@ export const DEFAULT_PRICING_TABLE = {
     // same per-token rate). Output price includes thinking tokens. $1/$5 is
     // introductory per ai.google.dev, verified 2026-10-08: Google lists
     // $2/$10 starting 2027-01-01, so bump these rates once that date passes.
+    // Cache read is likewise introductory: $0.10 through 2026-12-31, $0.20 after.
     'gemini-robotics-er-2-preview': {
+        inputPerMTok: 1,
+        cacheReadPerMTok: 0.1,
+        outputPerMTok: 5,
+        thinkingPerMTok: 5,
+        contextWindow: 131_072,
+    },
+    // Live API streaming endpoint; same introductory $1/$5 -> $2/$10 on
+    // 2027-01-01. Google lists no context caching for it.
+    'gemini-robotics-er-2-streaming-preview': {
         inputPerMTok: 1,
         outputPerMTok: 5,
         thinkingPerMTok: 5,
@@ -682,10 +780,10 @@ export const DEFAULT_PRICING_TABLE = {
     // modeled here: no cacheCreationPerMTok is set, so any cache-write tokens
     // that extractOpenAITokens() reports (prompt_tokens_details.cache_write_tokens)
     // are billed at $0.
-    // The GPT-5.5/5.6 pages (verified 2026-10-08) say the >272k tier is "2x
-    // input and 1.5x output" and are silent on cached input, so no
-    // tierCacheReadPerMTok is set for them (unlike GPT-6, which states 2x
-    // "input and cache rates"). Cached input above 272k bills at the base rate.
+    // The GPT-5.5/5.6 model pages say the >272k tier is "2x input and 1.5x
+    // output" and are silent on cached input, but the pricing page's "Long
+    // context cached input" column (verified 2026-10-26) lists 2x the base
+    // cached rate for GPT-5.4/5.5/5.6, so tierCacheReadPerMTok is set to match.
     // $4/$20 is promotional pricing per OpenAI's own model page: "a 20%
     // reduction in input pricing and a 33% reduction in output pricing"
     // versus GPT-5.5 ($5/$30), "available at least through November 21,
@@ -698,6 +796,7 @@ export const DEFAULT_PRICING_TABLE = {
         cacheReadPerMTok: 0.4,
         contextWindow: 1_050_000,
         tierThreshold: 272_000,
+        tierCacheReadPerMTok: 0.8,
         tierInputPerMTok: 8,
         tierOutputPerMTok: 30,
     },
@@ -707,6 +806,7 @@ export const DEFAULT_PRICING_TABLE = {
         cacheReadPerMTok: 0.2,
         contextWindow: 1_050_000,
         tierThreshold: 272_000,
+        tierCacheReadPerMTok: 0.4,
         tierInputPerMTok: 4,
         tierOutputPerMTok: 18,
     },
@@ -716,6 +816,7 @@ export const DEFAULT_PRICING_TABLE = {
         cacheReadPerMTok: 0.02,
         contextWindow: 1_050_000,
         tierThreshold: 272_000,
+        tierCacheReadPerMTok: 0.04,
         tierInputPerMTok: 0.4,
         tierOutputPerMTok: 1.8,
     },
@@ -761,6 +862,7 @@ export const DEFAULT_PRICING_TABLE = {
         // that was wrong on three counts: mode, threshold (270k vs 272k), and
         // the missing output multiplier.
         tierThreshold: 272_000,
+        tierCacheReadPerMTok: 1,
         tierInputPerMTok: 10,
         tierOutputPerMTok: 45,
     },
@@ -770,6 +872,7 @@ export const DEFAULT_PRICING_TABLE = {
         cacheReadPerMTok: 0.25,
         contextWindow: 1_050_000,
         tierThreshold: 272_000,
+        tierCacheReadPerMTok: 0.5,
         tierInputPerMTok: 5,
         tierOutputPerMTok: 22.5,
     },
@@ -989,10 +1092,11 @@ export const DEFAULT_PRICING_TABLE = {
     // Cache pricing for these models is published on the Bedrock pricing
     // page — resolved by matching
     // each model's known direct-API cache-write/-read multipliers (1.25x /
-    // 0.1x of input), which line up exactly with AWS's own numbers. AWS
+    // 0.1x of input), which line up exactly with AWS's own numbers (except the
+    // 0.05x Sonnet/Opus 5.5 and 0.025x Fable/Mythos 5.1 cache reads). AWS
     // publishes BOTH a 5-minute and a 1-hour cache-write rate; cacheCreationPerMTok
-    // here is the 5-minute rate (matches the direct-API default TTL) — there's
-    // no field for the 1-hour rate, which would need a schema change to model.
+    // here is the 5-minute rate (matches the direct-API default TTL) and
+    // cacheCreation1hPerMTok the 1-hour rate (2x input) where AWS lists one.
     // $2/$10 is Anthropic's standing Sonnet 5 price on Bedrock too — not
     // introductory. See the direct-API claude-sonnet-5 comment above.
     //
@@ -1019,16 +1123,19 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 11,
         cacheReadPerMTok: 0.22,
         cacheCreationPerMTok: 2.75,
+        cacheCreation1hPerMTok: 4.4,
         contextWindow: 1_000_000,
     },
     // Sonnet 5.5 and Opus 5.5 geo rates are read directly from AWS's pricing
-    // page (2026-10-07), not derived from the 1.1x multiplier. Model IDs
+    // page (2026-10-07), not derived from the 1.1x multiplier.
+    // Sonnet 5.5 cache reads are 0.05x input (cut 50% effective 2026-10-07). Model IDs
     // confirmed against AWS's model cards (2026-10-07).
     'anthropic.claude-sonnet-5-5': {
         inputPerMTok: 2.2,
         outputPerMTok: 11,
-        cacheReadPerMTok: 0.22,
+        cacheReadPerMTok: 0.11,
         cacheCreationPerMTok: 2.75,
+        cacheCreation1hPerMTok: 4.4,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-opus-5-5': {
@@ -1036,6 +1143,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 22,
         cacheReadPerMTok: 0.22,
         cacheCreationPerMTok: 5.5,
+        cacheCreation1hPerMTok: 8.8,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-fable-5': {
@@ -1043,6 +1151,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 55,
         cacheReadPerMTok: 1.1,
         cacheCreationPerMTok: 13.75,
+        cacheCreation1hPerMTok: 22.0,
         contextWindow: 1_000_000,
     },
     // Geo rate read from AWS's pricing page (2026-10-07).
@@ -1051,6 +1160,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 55,
         cacheReadPerMTok: 0.275,
         cacheCreationPerMTok: 13.75,
+        cacheCreation1hPerMTok: 22.0,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-opus-5': {
@@ -1058,6 +1168,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 27.5,
         cacheReadPerMTok: 0.55,
         cacheCreationPerMTok: 6.875,
+        cacheCreation1hPerMTok: 11.0,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-opus-4-8': {
@@ -1065,6 +1176,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 27.5,
         cacheReadPerMTok: 0.55,
         cacheCreationPerMTok: 6.875,
+        cacheCreation1hPerMTok: 11.0,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-opus-4-7': {
@@ -1072,6 +1184,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 27.5,
         cacheReadPerMTok: 0.55,
         cacheCreationPerMTok: 6.875,
+        cacheCreation1hPerMTok: 11.0,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-sonnet-4-6': {
@@ -1079,6 +1192,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 16.5,
         cacheReadPerMTok: 0.33,
         cacheCreationPerMTok: 4.125,
+        cacheCreation1hPerMTok: 6.6,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-haiku-4-5-20251001-v1:0': {
@@ -1086,34 +1200,41 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 5.5,
         cacheReadPerMTok: 0.11,
         cacheCreationPerMTok: 1.375,
+        cacheCreation1hPerMTok: 2.2,
         contextWindow: 200_000,
     },
     // Haiku 5.5: IDs from its AWS model card, rates from the AWS pricing page (2026-10-07).
     // The page's "Long Context Pricing" table lists 5x input/output/cache-read; it
-    // states no threshold, so the first-party 100k flat threshold is assumed. Cache
-    // writes stay at the base rate, as on the first-party entry.
+    // states no threshold, so the first-party 100k flat threshold is assumed. Long-
+    // context cache writes are also 5x the base rate (tierCacheCreationPerMTok).
     'anthropic.claude-haiku-5-5': {
         inputPerMTok: 0.11,
         outputPerMTok: 0.55,
         thinkingPerMTok: 0.55,
         cacheReadPerMTok: 0.011,
         cacheCreationPerMTok: 0.1375,
+        cacheCreation1hPerMTok: 0.22,
         contextWindow: 1_000_000,
         tierThreshold: 100_000,
         tierInputPerMTok: 0.55,
         tierOutputPerMTok: 2.75,
         tierThinkingPerMTok: 2.75,
         tierCacheReadPerMTok: 0.055,
+        tierCacheCreationPerMTok: 0.6875,
+        tierCacheCreation1hPerMTok: 1.1,
     },
     // Older and Mythos-line Claude models, geo/in-region rates read from AWS's
     // pricing page (2026-10-07); IDs confirmed against AWS model cards except
     // Mythos Preview, Opus 4, 3.7 Sonnet and 3 Sonnet (no card; AWS still prices them).
-    // Mythos 5 is bedrock-mantle (Messages API) only, with no Geo or Global profile.
+    // Mythos 5 is bedrock-mantle (Messages API) only; AWS's pricing page lists a
+    // Geo row for it (these rates, 2026-10-26) and no Global row. Its us./eu./au./
+    // jp. aliases are unverified, like the other regional Claude aliases.
     'anthropic.claude-mythos-5': {
         inputPerMTok: 11,
         outputPerMTok: 55,
         cacheReadPerMTok: 1.1,
         cacheCreationPerMTok: 13.75,
+        cacheCreation1hPerMTok: 22.0,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-mythos-5-1': {
@@ -1121,6 +1242,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 55,
         cacheReadPerMTok: 0.275,
         cacheCreationPerMTok: 13.75,
+        cacheCreation1hPerMTok: 22.0,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-mythos-preview': {
@@ -1128,6 +1250,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 137.5,
         cacheReadPerMTok: 2.75,
         cacheCreationPerMTok: 34.375,
+        cacheCreation1hPerMTok: 55.0,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-opus-4-6-v1': {
@@ -1135,6 +1258,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 27.5,
         cacheReadPerMTok: 0.55,
         cacheCreationPerMTok: 6.875,
+        cacheCreation1hPerMTok: 11.0,
         contextWindow: 1_000_000,
     },
     'anthropic.claude-opus-4-5-20251101-v1:0': {
@@ -1142,6 +1266,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 27.5,
         cacheReadPerMTok: 0.55,
         cacheCreationPerMTok: 6.875,
+        cacheCreation1hPerMTok: 11.0,
         contextWindow: 200_000,
     },
     'anthropic.claude-sonnet-4-5-20250929-v1:0': {
@@ -1149,6 +1274,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 16.5,
         cacheReadPerMTok: 0.33,
         cacheCreationPerMTok: 4.125,
+        cacheCreation1hPerMTok: 6.6,
         contextWindow: 200_000,
     },
     'anthropic.claude-sonnet-4-20250514-v1:0': {
@@ -1198,13 +1324,15 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 10,
         cacheReadPerMTok: 0.2,
         cacheCreationPerMTok: 2.5,
+        cacheCreation1hPerMTok: 4.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-sonnet-5-5': {
         inputPerMTok: 2,
         outputPerMTok: 10,
-        cacheReadPerMTok: 0.2,
+        cacheReadPerMTok: 0.1,
         cacheCreationPerMTok: 2.5,
+        cacheCreation1hPerMTok: 4.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-opus-5-5': {
@@ -1212,6 +1340,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 20,
         cacheReadPerMTok: 0.2,
         cacheCreationPerMTok: 5,
+        cacheCreation1hPerMTok: 8.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-fable-5': {
@@ -1219,6 +1348,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 50,
         cacheReadPerMTok: 1,
         cacheCreationPerMTok: 12.5,
+        cacheCreation1hPerMTok: 20.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-opus-5': {
@@ -1226,6 +1356,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-opus-4-8': {
@@ -1233,6 +1364,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-opus-4-7': {
@@ -1240,6 +1372,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-sonnet-4-6': {
@@ -1247,6 +1380,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 15,
         cacheReadPerMTok: 0.3,
         cacheCreationPerMTok: 3.75,
+        cacheCreation1hPerMTok: 6.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-haiku-4-5-20251001-v1:0': {
@@ -1254,6 +1388,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 5,
         cacheReadPerMTok: 0.1,
         cacheCreationPerMTok: 1.25,
+        cacheCreation1hPerMTok: 2.0,
         contextWindow: 200_000,
     },
     'global.anthropic.claude-opus-4-6-v1': {
@@ -1261,6 +1396,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-opus-4-5-20251101-v1:0': {
@@ -1268,6 +1404,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 25,
         cacheReadPerMTok: 0.5,
         cacheCreationPerMTok: 6.25,
+        cacheCreation1hPerMTok: 10.0,
         contextWindow: 200_000,
     },
     'global.anthropic.claude-sonnet-4-5-20250929-v1:0': {
@@ -1275,6 +1412,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 15,
         cacheReadPerMTok: 0.3,
         cacheCreationPerMTok: 3.75,
+        cacheCreation1hPerMTok: 6.0,
         contextWindow: 200_000,
     },
     'global.anthropic.claude-haiku-5-5': {
@@ -1283,21 +1421,24 @@ export const DEFAULT_PRICING_TABLE = {
         thinkingPerMTok: 0.5,
         cacheReadPerMTok: 0.01,
         cacheCreationPerMTok: 0.125,
+        cacheCreation1hPerMTok: 0.2,
         contextWindow: 1_000_000,
         tierThreshold: 100_000,
         tierInputPerMTok: 0.5,
         tierOutputPerMTok: 2.5,
         tierThinkingPerMTok: 2.5,
         tierCacheReadPerMTok: 0.05,
+        tierCacheCreationPerMTok: 0.625,
+        tierCacheCreation1hPerMTok: 1.0,
     },
-    // Fable 5.1 and Mythos 5.1: AWS model cards confirm these global IDs exist, but
-    // the pricing page lists no Global-tier row. Rates are DERIVED from the geo row
-    // via AWS's flat 1.1x geo premium; replace with published rates when AWS lists them.
+    // Fable 5.1 and Mythos 5.1: global rates read from AWS's pricing page (2026-10-26);
+    // they match the geo row divided by AWS's flat 1.1x geo premium.
     'global.anthropic.claude-fable-5-1': {
         inputPerMTok: 10,
         outputPerMTok: 50,
         cacheReadPerMTok: 0.25,
         cacheCreationPerMTok: 12.5,
+        cacheCreation1hPerMTok: 20.0,
         contextWindow: 1_000_000,
     },
     'global.anthropic.claude-mythos-5-1': {
@@ -1305,6 +1446,7 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 50,
         cacheReadPerMTok: 0.25,
         cacheCreationPerMTok: 12.5,
+        cacheCreation1hPerMTok: 20.0,
         contextWindow: 1_000_000,
     },
     // Sonnet 4 is priced identically in the Geo and Global tiers on the AWS page.
@@ -1352,6 +1494,29 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 75,
         contextWindow: 200_000,
     },
+    // Legacy Claude models AWS still prices on-demand (pricing page, 2026-10-26,
+    // no cache rates). Model IDs and context windows are the well-known Bedrock
+    // values, not confirmed against model cards.
+    'anthropic.claude-3-5-sonnet-20240620-v1:0': {
+        inputPerMTok: 3,
+        outputPerMTok: 15,
+        contextWindow: 200_000,
+    },
+    'anthropic.claude-v2:1': {
+        inputPerMTok: 8,
+        outputPerMTok: 24,
+        contextWindow: 200_000,
+    },
+    'anthropic.claude-v2': {
+        inputPerMTok: 8,
+        outputPerMTok: 24,
+        contextWindow: 100_000,
+    },
+    'anthropic.claude-instant-v1': {
+        inputPerMTok: 0.8,
+        outputPerMTok: 2.4,
+        contextWindow: 100_000,
+    },
     // Meta Llama via Bedrock
     // Llama 3 (both sizes) has an 8K context window, not 128K — 128K arrived
     // with Llama 3.1. 70B's rate was also stale: $0.99/$0.99 was actually
@@ -1499,20 +1664,18 @@ export const DEFAULT_PRICING_TABLE = {
     },
     // Mistral Large 4 (docs.mistral.ai/models/mistral-large-4-0, v26.10, public
     // preview since 2026-10-06; API names mistral-large-4 / mistral-large-4-0).
-    // Rates are the preview SALE price actually billed, verified 2026-10-07:
-    // $0.68 in / $2.09 out against a struck-through list of $1.36 / $4.18.
-    // TEMPORARY SALE PRICING, no published end date: re-verify at every pricing
-    // audit and revert to the list price ($1.36 / $4.18) once the sale ends,
-    // otherwise costs are under-reported by half. 'mistral-large-latest' above is
-    // deliberately unchanged: the docs don't say it now routes here.
+    // The preview sale ($0.68 / $2.09) is no longer shown on the docs page as of
+    // 2026-10-26, which lists only the $1.36 / $4.18 list price. Cached input
+    // ($0.14) is not modeled. 'mistral-large-latest' above is deliberately
+    // unchanged: the docs don't say it now routes here.
     'mistral-large-4': {
-        inputPerMTok: 0.68,
-        outputPerMTok: 2.09,
+        inputPerMTok: 1.36,
+        outputPerMTok: 4.18,
         contextWindow: 1_000_000,
     },
     'mistral-large-4-0': {
-        inputPerMTok: 0.68,
-        outputPerMTok: 2.09,
+        inputPerMTok: 1.36,
+        outputPerMTok: 4.18,
         contextWindow: 1_000_000,
     },
     'mistral-small-latest': {
@@ -1603,6 +1766,16 @@ export const DEFAULT_PRICING_TABLE = {
         outputPerMTok: 0.6,
         contextWindow: 128_000,
     },
+    // Command A+ is the only Command A model with a published API rate
+    // (cohere.com/pricing, 128K context and 64K max output per
+    // docs.cohere.com/docs/models, verified 2026-10-26). command-a-03-2025,
+    // -reasoning, -vision and -translate have no published rates and are
+    // deliberately absent.
+    'command-a-plus-05-2026': {
+        inputPerMTok: 2.5,
+        outputPerMTok: 10,
+        contextWindow: 128_000,
+    },
     'command-r7b-12-2024': {
         inputPerMTok: 0.0375,
         outputPerMTok: 0.15,
@@ -1749,6 +1922,8 @@ export const DEFAULT_PRICING_TABLE = {
     // split is only in response headers, so 1h writes under-price by $3.00/MTok.
     // GitHub's retirement table removes Kimi K2.7 Code from Copilot on 2026-10-02,
     // but Moonshot's direct page still lists it, so the entry stays.
+    // Moonshot's banner (2026-10-26): the K2 series closes to new users on
+    // 2026-10-31 and all K2 models retire by the end of 2026-12.
     'kimi-k2.7-code': {
         inputPerMTok: 0.95,
         outputPerMTok: 4,
@@ -1762,6 +1937,22 @@ export const DEFAULT_PRICING_TABLE = {
         cacheCreationPerMTok: 3,
         contextWindow: 1_048_576,
     },
+    // K3.1 Pro (launched 2026-10; verified 2026-10-26 on the same page). The 1h
+    // cache-write rate ($4.00, $8.00 highspeed) is unmodeled, as for K3.
+    'kimi-k3.1-pro': {
+        inputPerMTok: 2,
+        outputPerMTok: 10,
+        cacheReadPerMTok: 0.2,
+        cacheCreationPerMTok: 2,
+        contextWindow: 1_048_576,
+    },
+    'kimi-k3.1-pro-highspeed': {
+        inputPerMTok: 4,
+        outputPerMTok: 20,
+        cacheReadPerMTok: 0.2,
+        cacheCreationPerMTok: 4,
+        contextWindow: 1_048_576,
+    },
     // kimi-k2.7-code-highspeed and kimi-k2.6 verified 2026-10-07 against the
     // same page (column headers: cache hit / cache miss / output).
     'kimi-k2.7-code-highspeed': {