@oh-my-pi/pi-catalog 18.4.9 → 18.4.11

package/CHANGELOG.md

@@ -2,6 +2,22 @@
 
 ## [Unreleased]
 
+## [18.4.11] - 2026-10-02
+
+### Fixed
+
+- Fixed new Fireworks sessions failing on the first turn by updating the default model to `kimi-k3`, which is currently supported by Fireworks.
+
+## [18.4.10] - 2026-10-02
+
+### Fixed
+
+- Fixed MiniMax Token Plan (`minimax-code`, `minimax-code-cn`) usage showing as free; turns now show the pay-as-you-go equivalent cost, with MiniMax-M3.1-Flash-Preview estimated at the MiniMax-M3 rate since it has no published price ([#13695](https://github.com/can1357/oh-my-pi/pull/13695) by [@eggpeat](https://github.com/eggpeat))
+- Fixed namespaced LiteLLM models such as `azure/gpt-5.6-sol-pro` showing raw IDs instead of catalog display names when the proxy supplies no friendly name ([#13964](https://github.com/can1357/oh-my-pi/pull/13964) by [@gabrielrinaldi](https://github.com/gabrielrinaldi)).
+- Bundled prompt-cache lifetimes are recomputed from current policy instead of being inherited from previous generated models ([#13966](https://github.com/can1357/oh-my-pi/pull/13966) by [@anatoli-tsinovoy](https://github.com/anatoli-tsinovoy)).
+- Fixed HTTP 400 `messages.N.output_config: Extra inputs are not permitted` on later turns with Claude Sonnet 5.5, Opus 5, Opus 5.5, and Fable 5.1 on Google Vertex AI ([#13994](https://github.com/can1357/oh-my-pi/issues/13994))
+- Fixed Claude Opus 5.5 conversations failing with a "bound to a different conversation" 400 after the system prompt changed. Opus 5.5 now gets Sonnet 5.5's prefix-bound thinking handling on every provider, and Bedrock asks the server to drop stale signed thinking instead of rejecting the request ([#14019](https://github.com/can1357/oh-my-pi/pull/14019) by [@nick-maderight](https://github.com/nick-maderight))
+
 ## [18.4.9] - 2026-10-01
 
 ### Fixed

package/dist/types/provider-models/openai-compat.d.ts

@@ -159,12 +159,6 @@ export interface XaiModelManagerConfig {
     baseUrl?: string;
     fetch?: FetchImpl;
 }
-/**
- * Mirrors exact public-model prices onto matching SuperGrok catalog rows.
- * The >200K long-context tier itself is rule-owned (`classes/xai.kdl`
- * `long-context-cost` multiplier axis) and derives at build time.
- */
-export declare function applyXaiCatalogPricing(models: readonly ModelSpec[]): ModelSpec[];
 export declare function xaiModelManagerOptions(config?: XaiModelManagerConfig): ModelManagerOptions<"openai-responses">;
 export interface XaiOAuthModelManagerConfig {
     apiKey?: string;

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "18.4.9",
+	"version": "18.4.11",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
 	"keywords": [
 		"ai",
@@ -93,12 +93,12 @@
 		"gen:proto": "bun scripts/generate-protocols.ts"
 	},
 	"dependencies": {
-		"@oh-my-pi/omptype": "18.4.9",
-		"@oh-my-pi/pi-utils": "18.4.9"
+		"@oh-my-pi/omptype": "18.4.11",
+		"@oh-my-pi/pi-utils": "18.4.11"
 	},
 	"devDependencies": {
 		"@bgotink/kdl": "0.4.0",
-		"@oh-my-pi/pi-ai": "18.4.9",
+		"@oh-my-pi/pi-ai": "18.4.11",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/compat/rules/classes/anthropic.kdl

@@ -59,6 +59,13 @@ class "anthropic" {
 	// to be re-verified against the API before it inherits the restriction.
 	family "opus" {
 		revision ">=5.5 <6" {
+			// Opus 5.5 binds signed thinking to the exact preceding conversation, so a
+			// replay fails once the system prompt, tools or earlier messages change.
+			// Like the Sonnet 5.5 rule, this one spans every provider, so it changes
+			// pruning, implicit prompt refreshes and thinking replay everywhere. On
+			// Bedrock Converse it also sends `block_binding` with its beta.
+			// https://platform.claude.com/docs/en/build-with-claude/preserved-thinking
+			thinking-prefix-binding #true
 			supports-forced-tool-choice #false
 		}
 	}
@@ -84,9 +91,6 @@ class "anthropic" {
 				supports-mid-conversation-tool-changes #true
 				supports-turn-scoped-system #true
 			}
-			revision ">=5.1" {
-				supports-per-message-effort #true
-			}
 		}
 		family "mythos" {
 			revision ">=5" {
@@ -94,9 +98,6 @@ class "anthropic" {
 				supports-mid-conversation-tool-changes #true
 				supports-turn-scoped-system #true
 			}
-			revision ">=5.1" {
-				supports-per-message-effort #true
-			}
 		}
 		family "fable" {
 			revision ">=5.1" {
@@ -108,7 +109,6 @@ class "anthropic" {
 				supports-mid-conversation-system #true
 				supports-mid-conversation-tool-changes #true
 				supports-turn-scoped-system #true
-				supports-per-message-effort #true
 			}
 		}
 		family "opus" {
@@ -117,17 +117,33 @@ class "anthropic" {
 				supports-mid-conversation-tool-changes #true
 				supports-turn-scoped-system #true
 			}
-			revision ">=5" {
-				supports-per-message-effort #true
-			}
 		}
 	}
 	// Vertex rejects `thinking.block_binding` on Sonnet 5.5 with 400
-	// `thinking.adaptive.block_binding: Extra inputs are not permitted`.
+	// `thinking.adaptive.block_binding: Extra inputs are not permitted`, and
+	// per-message `output_config` with 400 `messages.N.output_config: Extra
+	// inputs are not permitted` (it takes no `output_config.effort` at all; see
+	// `supports-output-effort` in providers/google-vertex.kdl).
 	on "anthropic" "cloudflare-ai-gateway" {
+		family "fable" {
+			revision ">=5.1" {
+				supports-per-message-effort #true
+			}
+		}
+		family "mythos" {
+			revision ">=5.1" {
+				supports-per-message-effort #true
+			}
+		}
 		family "sonnet" {
 			revision ">=5.5" {
 				supports-thinking-binding-controls #true
+				supports-per-message-effort #true
+			}
+		}
+		family "opus" {
+			revision ">=5" {
+				supports-per-message-effort #true
 			}
 		}
 	}

package/src/compat/rules/providers/fireworks.kdl

@@ -1,7 +1,7 @@
 // Provider-wire compat for "fireworks"; regenerated from the frozen census using deployment contract selectors.
 
 provider "fireworks" priority=-1 {
-	default-model "kimi-k2.7-code"
+	default-model "kimi-k3"
 	env "FIREWORKS_API_KEY"
 	discovery label="Fireworks"
 

package/src/compat/rules/runtime/behavior.kdl

@@ -172,7 +172,7 @@ behavior {
 	// SuperGrok's roster interleaves media models the chat picker cannot
 	// serve; the dedicated tool surfaces route them instead.
 	exclude-models provider="xai-oauth" exact="grok-imagine-image-quality" exact="grok-imagine-image-2.0" \
-		exact="grok-imagine-video" exact="grok-imagine-video-1.5" prefix="grok-stt-" prefix="grok-voice-"
+		prefix="grok-imagine-video" prefix="grok-stt-" prefix="grok-voice-"
 	// Meta's /v1/models lists image generation and transcription SKUs beside
 	// the Muse Spark chat models.
 	exclude-models provider="meta" prefix="muse-image-" prefix="muse-voice-"
@@ -357,6 +357,16 @@ behavior {
 	pricing-peer provider="xai-oauth" peers="xai" {
 		alias "grok-4.20-multi-agent-0309" peer-id="grok-4.20-multi-agent-beta-latest"
 	}
+	// Token Plan rows report zero cost upstream; like the Antigravity and
+	// SuperGrok rules above, they show the pay-as-you-go equivalent (Credits
+	// overflow is billed at the PAYG list price). MiniMax-M3.1-Flash-Preview has
+	// no published price and borrows the MiniMax-M3 rate as an estimate.
+	pricing-peer provider="minimax-code" peers="minimax" {
+		alias "MiniMax-M3.1-Flash-Preview" peer-id="MiniMax-M3"
+	}
+	pricing-peer provider="minimax-code-cn" peers="minimax-cn" "minimax" {
+		alias "MiniMax-M3.1-Flash-Preview" peer-id="MiniMax-M3"
+	}
 	hosted-default provider="zai-search" model="glm-4.7"
 	hosted-default provider="synthetic-search" model="auto"
 }