pi-coralbricks-provider 1.0.27 → 1.0.28

package/README.md

@@ -85,7 +85,7 @@ pi
 
 | Model | Context | Vision | Reasoning | Input $/M | Cache Read $/M | Cache Write $/M | Output $/M |
 |-------|---------|--------|-----------|-----------|-----------------|------------------|------------|
-| DeepSeek V4.1 Flash | 1.0M | ✅ | ✅ | $0.30 | — | $0.09 | $1.20 |
+| DeepSeek V4.1 Flash | 1.0M | ✅ | ✅ | $0.01 | — | $0.09 | $1.20 |
 | GLM 5.3 | 1.0M | ❌ | ✅ | $1.12 | — | $1.68 | $4.40 |
 
 *Costs are per million tokens. Cache Read shows — because Coral bills cached input at **$0** on every model. Prompt tokens Coral has not cached yet are billed once at the Cache Write rate, in place of the Input rate. Prices subject to change — check [Coral's live catalog](https://www.coralbricks.ai/api/public/models).*

package/index.ts

@@ -197,6 +197,17 @@ function toNumber(value: unknown): number {
   return typeof value === "number" && Number.isFinite(value) ? value : 0;
 }
 
+/**
+ * Full price of a written prompt token. Since 2026-10-08 Coral states the cache write ON TOP
+ * of input (/v1/models cache_write_per_m, catalog cacheWritePerM), and a written token is
+ * billed input + that. pi prices a written token at cacheWrite alone, so add the two. 0 when
+ * the row carries no cache-write rate. Rounded so 1.12 + 0.56 is 1.68, not 1.6800000000000002.
+ */
+function writtenTokenCost(input: number, cacheWriteOnTop: unknown): number {
+  if (typeof cacheWriteOnTop !== "number" || !Number.isFinite(cacheWriteOnTop)) return 0;
+  return Math.round((input + cacheWriteOnTop) * 1e6) / 1e6;
+}
+
 function displayName(id: string): string {
   return id
     .split(/[-_/]/)
@@ -233,8 +244,9 @@ function baseCompat() {
 /**
  * Transform a Coral /v1/models row. Coral reports per-million USD pricing
  * (input_per_m / output_per_m / cached_input_per_m — cached reads are $0 on
- * every model — and cache_write_per_m, the rate for prompt tokens written to
- * the cache), context_length, and supports_image_input.
+ * every model — and cache_write_per_m, the cache write ON TOP of input for a
+ * prompt token written to the cache), context_length, and supports_image_input.
+ * pi's cacheWrite is the full price of a written token: input + cache_write_per_m.
  */
 function transformApiModel(apiModel: any): JsonModel | null {
   if (!apiModel?.id) return null;
@@ -248,7 +260,7 @@ function transformApiModel(apiModel: any): JsonModel | null {
       input: toNumber(pricing.input_per_m),
       output: toNumber(pricing.output_per_m),
       cacheRead: toNumber(pricing.cached_input_per_m),
-      cacheWrite: toNumber(pricing.cache_write_per_m),
+      cacheWrite: writtenTokenCost(toNumber(pricing.input_per_m), pricing.cache_write_per_m),
     },
     contextWindow: parseContextWindow(apiModel.context_length),
     maxTokens: DEFAULT_MAX_TOKENS[apiModel.id] ?? FALLBACK_MAX_TOKENS,
@@ -261,10 +273,10 @@ function transformApiModel(apiModel: any): JsonModel | null {
  * Transform a row from the public model catalog
  * (https://www.coralbricks.ai/api/public/models) — the no-auth mirror of the
  * gateway control plane. Fields: slug, name, contextWindow ("1M"),
- * inputPerM / outputPerM. Coral's own prices are authoritative here; the
- * parity/field vendor comparison fields are ignored. The catalog carries no
- * cache-write rate, so cacheWrite stays 0 and mergeWithEmbedded keeps the
- * embedded rate.
+ * inputPerM / outputPerM, and since 2026-10-08 cacheWritePerM (the cache write on
+ * top of input). Coral's own prices are authoritative here; the parity/field vendor
+ * comparison fields are ignored. A row without cacheWritePerM leaves cacheWrite 0 so
+ * mergeWithEmbedded keeps the embedded rate.
  */
 function transformCatalogModel(entry: any): JsonModel | null {
   if (!entry?.slug) return null;
@@ -277,7 +289,7 @@ function transformCatalogModel(entry: any): JsonModel | null {
       input: toNumber(entry.inputPerM),
       output: toNumber(entry.outputPerM),
       cacheRead: 0,
-      cacheWrite: 0,
+      cacheWrite: writtenTokenCost(toNumber(entry.inputPerM), entry.cacheWritePerM),
     },
     contextWindow: parseContextWindow(entry.contextWindow),
     maxTokens: DEFAULT_MAX_TOKENS[entry.slug] ?? FALLBACK_MAX_TOKENS,

package/models.json

@@ -8,7 +8,7 @@
       "image"
     ],
     "cost": {
-      "input": 0.3,
+      "input": 0.01,
       "output": 1.2,
       "cacheRead": 0,
       "cacheWrite": 0.09
@@ -30,7 +30,7 @@
       "image"
     ],
     "cost": {
-      "input": 0.3,
+      "input": 0.01,
       "output": 1.2,
       "cacheRead": 0,
       "cacheWrite": 0.09

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-coralbricks-provider",
-  "version": "1.0.27",
+  "version": "1.0.28",
   "description": "CoralBricks provider extension for pi - Access GLM 5.3 and DeepSeek V4.1 Flash (FP4 variants included) through the Coral Inference API",
   "author": "monotykamary",
   "homepage": "https://github.com/monotykamary/pi-coralbricks-provider#readme",

package/scripts/update-models.js

@@ -166,6 +166,13 @@ function saveJson(filePath, data) {
   console.log(`✓ Saved ${path.basename(filePath)}`);
 }
 
+// Coral states the cache write ON TOP of input (2026-10-08); pi's cacheWrite is the full
+// price of a written token, input + it. 0 when the row has no cache-write rate.
+function writtenTokenCost(input, cacheWriteOnTop) {
+  if (typeof cacheWriteOnTop !== 'number' || !Number.isFinite(cacheWriteOnTop)) return 0;
+  return Math.round((input + cacheWriteOnTop) * 1e6) / 1e6;
+}
+
 function toNumber(value) {
   return typeof value === 'number' && Number.isFinite(value) ? value : 0;
 }
@@ -274,7 +281,9 @@ function transformApiModel(apiModel, existingModelsMap) {
     if (typeof pricing.input_per_m === 'number') existing.cost.input = pricing.input_per_m;
     if (typeof pricing.output_per_m === 'number') existing.cost.output = pricing.output_per_m;
     existing.cost.cacheRead = toNumber(pricing.cached_input_per_m);
-    if (typeof pricing.cache_write_per_m === 'number') existing.cost.cacheWrite = pricing.cache_write_per_m;
+    if (typeof pricing.cache_write_per_m === 'number') {
+      existing.cost.cacheWrite = writtenTokenCost(existing.cost.input, pricing.cache_write_per_m);
+    }
     existing.input = input;
     return existing;
   }
@@ -289,7 +298,7 @@ function transformApiModel(apiModel, existingModelsMap) {
       input: toNumber(pricing.input_per_m),
       output: toNumber(pricing.output_per_m),
       cacheRead: toNumber(pricing.cached_input_per_m),
-      cacheWrite: toNumber(pricing.cache_write_per_m),
+      cacheWrite: writtenTokenCost(toNumber(pricing.input_per_m), pricing.cache_write_per_m),
     },
     contextWindow,
     maxTokens,
@@ -311,6 +320,9 @@ function transformCatalogModel(entry, existingModelsMap) {
     if (contextWindow) existing.contextWindow = contextWindow;
     if (typeof entry.inputPerM === 'number') existing.cost.input = entry.inputPerM;
     if (typeof entry.outputPerM === 'number') existing.cost.output = entry.outputPerM;
+    if (typeof entry.cacheWritePerM === 'number') {
+      existing.cost.cacheWrite = writtenTokenCost(existing.cost.input, entry.cacheWritePerM);
+    }
     return existing;
   }
 
@@ -323,7 +335,7 @@ function transformCatalogModel(entry, existingModelsMap) {
       input: toNumber(entry.inputPerM),
       output: toNumber(entry.outputPerM),
       cacheRead: 0,
-      cacheWrite: 0,
+      cacheWrite: writtenTokenCost(toNumber(entry.inputPerM), entry.cacheWritePerM),
     },
     contextWindow,
     maxTokens,