Integrate pricing with canonical model #6130

katzdave · 2025-12-15T23:47:24Z

Main changes are removing pricing.rs which maintains model pricing via a cached config file which is refreshed from open router. Use canonical models instead (same data source, but much higher hit rate on name matching).

Stop fetching all of the models/prices into the client, and trying to maintain that state. There was some UI buried in settings about when the last pricing data refresh occurred. Only do targeted fetches with (provider, model).

Result is no meaningful difference in behavior but many more models/providers supported with pricing (databricks!).

A slight loss on the freshness, but will continue to follow up on keeping canonical models as up to date as possible.

katzdave · 2025-12-16T19:33:21Z

ui/desktop/src/hooks/useCostTracking.ts

-    ) {
-      // Model/provider has changed, save the costs for the previous model
-      const prevKey = `${prevProviderRef.current}/${prevModelRef.current}`;
+    const handleModelChange = async () => {


This is client side only and doesn't get persisted...

Filed: #6141

Will fix in a followup.

Copilot

Pull request overview

This PR refactors the pricing system to use canonical models instead of maintaining a separate pricing cache. The key architectural change is moving from eagerly fetching and caching all model pricing data from OpenRouter to fetching pricing on-demand from the canonical model registry, which provides better model name matching (supporting providers like Databricks) at the cost of slightly less fresh data.

Removed the complex pricing.rs module with its disk cache and OpenRouter API integration
Simplified frontend pricing logic by removing the costDatabase session cache
Changed API from bulk fetch (configured_only flag) to targeted fetch (provider + model parameters)

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
ui/desktop/src/utils/pricing.ts	New simplified pricing utility using targeted backend fetch
ui/desktop/src/utils/costDatabase.ts	Removed complex session caching and OpenRouter model parsing logic
ui/desktop/src/hooks/useCostTracking.ts	Updated to use async fetch instead of sync cache lookup
ui/desktop/src/hooks/useAgent.ts	Removed pricing cache initialization
ui/desktop/src/components/settings/app/AppSettingsSection.tsx	Removed pricing refresh UI and status indicators
ui/desktop/src/components/bottom_menu/CostTracker.tsx	Simplified to fetch pricing directly without session cache
crates/goose/src/providers/pricing.rs	Removed entire OpenRouter pricing cache module
crates/goose/src/providers/mod.rs	Removed pricing module export
crates/goose/src/providers/canonical/mod.rs	Added helper to fetch canonical model with pricing
crates/goose-server/src/routes/config_management.rs	Changed pricing endpoint to targeted fetch using canonical models
crates/goose-server/src/commands/agent.rs	Removed pricing cache initialization on startup
crates/goose-cli/src/session/output.rs	Changed to sync pricing lookup from canonical models
crates/goose-cli/src/session/mod.rs	Removed pricing cache initialization

Copilot · 2025-12-16T19:38:57Z

ui/desktop/src/hooks/useCostTracking.ts

+    const handleModelChange = async () => {
+      if (
+        prevModelRef.current !== undefined &&
+        prevProviderRef.current !== undefined &&
+        (prevModelRef.current !== currentModel || prevProviderRef.current !== currentProvider)
+      ) {
+        // Model/provider has changed, save the costs for the previous model
+        const prevKey = `${prevProviderRef.current}/${prevModelRef.current}`;

-      // Get pricing info for the previous model
-      const prevCostInfo = getCostForModel(prevProviderRef.current, prevModelRef.current);
+        // Get pricing info for the previous model
+        const prevCostInfo = await fetchModelPricing(
+          prevProviderRef.current,
+          prevModelRef.current
+        );

-      if (prevCostInfo) {
-        const prevInputCost =
-          (sessionInputTokens || localInputTokens) * (prevCostInfo.input_token_cost || 0);
-        const prevOutputCost =
-          (sessionOutputTokens || localOutputTokens) * (prevCostInfo.output_token_cost || 0);
-        const prevTotalCost = prevInputCost + prevOutputCost;
+        if (prevCostInfo) {
+          const prevInputCost =
+            (sessionInputTokens || localInputTokens) * (prevCostInfo.input_token_cost || 0);
+          const prevOutputCost =
+            (sessionOutputTokens || localOutputTokens) * (prevCostInfo.output_token_cost || 0);
+          const prevTotalCost = prevInputCost + prevOutputCost;

-        // Save the accumulated costs for this model
-        setSessionCosts((prev) => ({
-          ...prev,
-          [prevKey]: {
-            inputTokens: sessionInputTokens || localInputTokens,
-            outputTokens: sessionOutputTokens || localOutputTokens,
-            totalCost: prevTotalCost,
-          },
-        }));
+          // Save the accumulated costs for this model
+          setSessionCosts((prev) => ({
+            ...prev,
+            [prevKey]: {
+              inputTokens: sessionInputTokens || localInputTokens,
+              outputTokens: sessionOutputTokens || localOutputTokens,
+              totalCost: prevTotalCost,
+            },
+          }));
+        }
+
+        console.log(
+          'Model changed from',
+          `${prevProviderRef.current}/${prevModelRef.current}`,
+          'to',
+          `${currentProvider}/${currentModel}`,
+          '- saved costs and restored session token counters'
+        );
      }

-      console.log(
-        'Model changed from',
-        `${prevProviderRef.current}/${prevModelRef.current}`,
-        'to',
-        `${currentProvider}/${currentModel}`,
-        '- saved costs and restored session token counters'
-      );
-    }
+      prevModelRef.current = currentModel || undefined;
+      prevProviderRef.current = currentProvider || undefined;
+    };

-    prevModelRef.current = currentModel || undefined;
-    prevProviderRef.current = currentProvider || undefined;
+    handleModelChange();


This async function inside useEffect doesn't properly handle race conditions or cleanup. If the effect runs again while a previous fetchModelPricing call is still pending, both calls will complete and potentially update state out of order. Consider using an abort controller or a flag to cancel or ignore stale requests.

Copilot · 2025-12-16T19:38:57Z

crates/goose-server/src/routes/config_management.rs

 pub async fn get_pricing(
    Json(query): Json<PricingQuery>,
 ) -> Result<Json<PricingResponse>, StatusCode> {
-    let configured_only = query.configured_only;
-
-    // If refresh requested (configured_only = false), refresh the cache
-    if !configured_only {
-        if let Err(e) = refresh_pricing().await {
-            tracing::error!("Failed to refresh pricing data: {}", e);
-        }
-    }
+    let canonical_model =
+        maybe_get_canonical_model(&query.provider, &query.model).ok_or(StatusCode::NOT_FOUND)?;

    let mut pricing_data = Vec::new();

-    if !configured_only {
-        // Get ALL pricing data from the cache
-        let all_pricing = get_all_pricing().await;
-
-        for (provider, models) in all_pricing {
-            for (model, pricing) in models {
-                pricing_data.push(PricingData {
-                    provider: provider.clone(),
-                    model: model.clone(),
-                    input_token_cost: pricing.input_cost,
-                    output_token_cost: pricing.output_cost,
-                    currency: "$".to_string(),
-                    context_length: pricing.context_length,
-                });
-            }
-        }
-    } else {
-        for (metadata, provider_type) in get_providers().await {
-            // Skip unconfigured providers if filtering
-            if !check_provider_configured(&metadata, provider_type) {
-                continue;
-            }
-
-            for model_info in &metadata.known_models {
-                // Handle OpenRouter models specially - they store full provider/model names
-                let (lookup_provider, lookup_model) = if metadata.name == "openrouter" {
-                    // For OpenRouter, parse the model name to extract real provider/model
-                    if let Some((provider, model)) = parse_model_id(&model_info.name) {
-                        (provider, model)
-                    } else {
-                        // Fallback if parsing fails
-                        (metadata.name.clone(), model_info.name.clone())
-                    }
-                } else {
-                    // For other providers, use names as-is
-                    (metadata.name.clone(), model_info.name.clone())
-                };
-
-                // Only get pricing from OpenRouter cache
-                if let Some(pricing) = get_model_pricing(&lookup_provider, &lookup_model).await {
-                    pricing_data.push(PricingData {
-                        provider: metadata.name.clone(),
-                        model: model_info.name.clone(),
-                        input_token_cost: pricing.input_cost,
-                        output_token_cost: pricing.output_cost,
-                        currency: "$".to_string(),
-                        context_length: pricing.context_length,
-                    });
-                }
-                // No fallback to hardcoded prices
-            }
-        }
+    if let (Some(input_cost), Some(output_cost)) = (
+        canonical_model.pricing.prompt,
+        canonical_model.pricing.completion,
+    ) {
+        pricing_data.push(PricingData {
+            provider: query.provider.clone(),
+            model: query.model.clone(),
+            input_token_cost: input_cost,
+            output_token_cost: output_cost,
+            currency: "$".to_string(),
+            context_length: Some(canonical_model.context_length as u32),
+        });
    }

-    tracing::debug!(
-        "Returning pricing for {} models{}",
-        pricing_data.len(),
-        if configured_only {
-            " (configured providers only)"
-        } else {
-            " (all cached models)"
-        }
-    );
-
    Ok(Json(PricingResponse {
        pricing: pricing_data,
-        source: "openrouter".to_string(),
+        source: "canonical".to_string(),
    }))
 }


The API can return 200 OK with an empty pricing array when a canonical model exists but has no pricing data (prompt or completion is None). This differs from returning 404 when the model isn't found. Consider returning 404 or a different status when pricing is unavailable to distinguish between "model not found" and "model found but no pricing".

alexhancock

looks awesome - added one comment / question

alexhancock · 2025-12-16T20:39:47Z

ui/desktop/src/utils/pricing.ts

@@ -0,0 +1,60 @@
+import { getApiUrl } from '../config';
+
+export interface ModelCostInfo {


should we make this a type in the openapi api and get it from there instead? that way we could use the generated SDK for fetching it too.

can't we remove this type now and use PricingData? https://github.com/block/goose/pull/6130/files#diff-270e779202976f7f9b473ced6ebce8debeba7c7d7e5085991d08d5a363f1f40fR530

that's what I was after

after that, LGTM for a merge!

Ah misunderstood slightly. Done

…icing * 'main' of github.com:block/goose: (35 commits) docs: skills (#6062) fix: add conditional configuration for GOOSE_BIN_DIR in PATH (#5940) Update dependencies to help in Fedora packaging (#5835) fix: make goose reviewer less bad (#6154) docs: create/edit recipe button (#6145) fix(google): Fix 400 Bad Request error with Gemini 3 thought signatures (#6035) fix: we don't need to warn about tool count when in code mode (#6149) deps: upgrade agent-client-protocol to 0.9.0 (#6109) fix(providers): fix for gemini-cli on windows to work around cmd's multiline prompt limitations #5911 (#5966) More slash commands (#5858) fix: MCP UI not rendering due to CallToolResult structure change (#6143) fix: display shell output as static text instead of spinner (#6041) fix : Custom providers with empty API keys show as configured in desktop (#6105) Add .agents/skills and ~/.config/agent/skills to skills discovery paths (#6139) fix: use instructions for system prompt and prompt for user message in subagents (#6121) Fix compaction loop for small models or large input (#5803) feat: Centralize theme management with ThemeContext (#6137) OpenRouter & Xai streaming (#5873) fix: resolve mcp-hermit cleanup path expansion issue (#5953) feat: add goose PR reviewer workflow (#6124) ...

Copilot

Pull request overview

Copilot reviewed 17 out of 17 changed files in this pull request and generated 3 comments.

Copilot · 2025-12-18T16:23:55Z

crates/goose-server/src/routes/config_management.rs

+        pricing_data.push(PricingData {
+            provider: query.provider.clone(),
+            model: query.model.clone(),
+            input_token_cost: input_cost,
+            output_token_cost: output_cost,
+            currency: "$".to_string(),
+            context_length: Some(canonical_model.context_length as u32),


The cast from usize to u32 could overflow on 64-bit systems if context_length exceeds u32::MAX. Consider using try_into() with proper error handling or validate that the value fits within u32 range.

Suggested change

pricing_data.push(PricingData {

provider: query.provider.clone(),

model: query.model.clone(),

input_token_cost: input_cost,

output_token_cost: output_cost,

currency: "$".to_string(),

context_length: Some(canonical_model.context_length as u32),

let context_length = u32::try_from(canonical_model.context_length)

.map_err(|_| StatusCode::INTERNAL_SERVER_ERROR)?;

pricing_data.push(PricingData {

provider: query.provider.clone(),

model: query.model.clone(),

input_token_cost: input_cost,

output_token_cost: output_cost,

currency: "$".to_string(),

context_length: Some(context_length),

Copilot · 2025-12-18T16:23:55Z

crates/goose-server/src/routes/config_management.rs

+    if let (Some(input_cost), Some(output_cost)) = (
+        canonical_model.pricing.prompt,
+        canonical_model.pricing.completion,
+    ) {
+        pricing_data.push(PricingData {
+            provider: query.provider.clone(),
+            model: query.model.clone(),
+            input_token_cost: input_cost,
+            output_token_cost: output_cost,
+            currency: "$".to_string(),
+            context_length: Some(canonical_model.context_length as u32),
+        });
    }


When a model is found but has no pricing data (prompt or completion is None), the endpoint returns a 200 OK with an empty pricing array. The frontend at line 20 of pricing.ts expects pricing?.[0] ?? null, which correctly handles this. However, returning 404 NOT_FOUND when pricing data is unavailable would be more semantically correct and make error handling clearer.

Copilot · 2025-12-18T16:23:56Z

crates/goose/src/providers/canonical/mod.rs

+pub fn maybe_get_canonical_model(provider: &str, model: &str) -> Option<CanonicalModel> {
+    let registry = CanonicalModelRegistry::bundled().ok()?;
+    let canonical_id = map_to_canonical_model(provider, model, registry)?;
+    registry.get(&canonical_id).cloned()


The function calls CanonicalModelRegistry::bundled() twice - once on line 25 and implicitly again on line 27 when using the result. The registry should be stored in a variable after the first call to avoid the potential overhead of multiple lazy initialization checks and error conversions.

* 'main' of github.com:block/goose: (28 commits) Clean PR preview sites from gh-pages branch history (#6161) fix: make goose reviewer less sycophantic (#6171) revert /reply to previous behavior (replacing session history) when full conversation provided (#6058) chore: manually update version (#6166) Integrate pricing with canonical model (#6130) Regenerate canonical models when release branch is created. (#6127) fix: use correct parameter name in read_module handler (#6148) docs: blog for code mode MCP (#6126) test: add ACP integration test (#6150) docs: auto download updates (#6163) fix: respect default_enabled value of platform extensions (#6159) docs: skills (#6062) fix: add conditional configuration for GOOSE_BIN_DIR in PATH (#5940) Update dependencies to help in Fedora packaging (#5835) fix: make goose reviewer less bad (#6154) docs: create/edit recipe button (#6145) fix(google): Fix 400 Bad Request error with Gemini 3 thought signatures (#6035) fix: we don't need to warn about tool count when in code mode (#6149) deps: upgrade agent-client-protocol to 0.9.0 (#6109) fix(providers): fix for gemini-cli on windows to work around cmd's multiline prompt limitations #5911 (#5966) ... # Conflicts: # ui/desktop/src/api/sdk.gen.ts # ui/desktop/src/hooks/useAgent.ts

katzdave added 6 commits December 15, 2025 17:14

first pass

4d28872

cleanup more dead code

f4b8859

further cleanup

96cb5ec

massive UI pricing code dleetion

8234a5f

more dl

3aee1c6

fmt

77bd96e

katzdave changed the title ~~Massively simplify pricing code~~ Integrate pricing with canonical model Dec 16, 2025

katzdave commented Dec 16, 2025

View reviewed changes

Cleanup noisy comments

3584478

katzdave marked this pull request as ready for review December 16, 2025 19:34

Copilot AI review requested due to automatic review settings December 16, 2025 19:34

Copilot started reviewing on behalf of katzdave December 16, 2025 19:35 View session

Copilot AI reviewed Dec 16, 2025

View reviewed changes

alexhancock approved these changes Dec 16, 2025

View reviewed changes

katzdave added 4 commits December 17, 2025 14:21

use openapi

dec46a8

rm modelcost info

b084dda

rm free provider check

df6313d

Copilot AI review requested due to automatic review settings December 18, 2025 16:19

Copilot started reviewing on behalf of katzdave December 18, 2025 16:19 View session

Copilot AI reviewed Dec 18, 2025

View reviewed changes

katzdave merged commit 6a0b8c2 into main Dec 18, 2025
24 checks passed

katzdave deleted the dkatz/canonical-pricing branch December 18, 2025 16:44

		@@ -0,0 +1,60 @@
		import { getApiUrl } from '../config';

		export interface ModelCostInfo {

Integrate pricing with canonical model #6130

Integrate pricing with canonical model #6130

Uh oh!

Conversation

katzdave commented Dec 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

katzdave Dec 16, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Dec 16, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 16, 2025

Choose a reason for hiding this comment

Uh oh!

alexhancock left a comment

Choose a reason for hiding this comment

Uh oh!

alexhancock Dec 16, 2025

Choose a reason for hiding this comment

Uh oh!

katzdave Dec 17, 2025

Choose a reason for hiding this comment

Uh oh!

alexhancock Dec 17, 2025

Choose a reason for hiding this comment

Uh oh!

katzdave Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

katzdave commented Dec 15, 2025 •

edited

Loading