@ottocode/sdk 0.1.197 → 0.1.199

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@ottocode/sdk",
-	"version": "0.1.197",
+	"version": "0.1.199",
 	"description": "AI agent SDK for building intelligent assistants - tree-shakable and comprehensive",
 	"author": "nitishxyz",
 	"license": "MIT",

package/src/core/src/tools/builtin/patch/apply.ts

@@ -435,6 +435,7 @@ function applyHunkToLines(
 	let matchIndex = hasExpected
 		? findSubsequence(lines, expected, Math.max(0, initialHint - 3), useFuzzy)
 		: -1;
+	let matchedExpected = expected;
 
 	if (hasExpected && matchIndex === -1) {
 		matchIndex = findSubsequence(lines, expected, 0, useFuzzy);
@@ -447,11 +448,24 @@ function applyHunkToLines(
 		if (!allContextPresent) {
 			matchIndex = -1;
 		} else {
-			const expectedWithoutMissingRemovals = expected.filter((line) =>
-				lineExists(lines, line, useFuzzy),
-			);
-			const minRequired = Math.max(2, Math.ceil(expected.length / 2));
-			if (expectedWithoutMissingRemovals.length >= minRequired) {
+			const expectedWithoutMissingRemovals = hunk.lines
+				.filter((line) => {
+					if (line.kind === 'add') return false;
+					if (line.kind === 'remove') {
+						return lineExists(lines, line.content, useFuzzy);
+					}
+					return true;
+				})
+				.map((line) => line.content);
+			const includedRemovalCount = hunk.lines.filter(
+				(line) =>
+					line.kind === 'remove' && lineExists(lines, line.content, useFuzzy),
+			).length;
+			const minRequired = Math.max(contextLines.length, 2);
+			if (
+				includedRemovalCount > 0 &&
+				expectedWithoutMissingRemovals.length >= minRequired
+			) {
 				matchIndex = findSubsequence(
 					lines,
 					expectedWithoutMissingRemovals,
@@ -466,6 +480,63 @@ function applyHunkToLines(
 						useFuzzy,
 					);
 				}
+				if (matchIndex !== -1) {
+					matchedExpected = expectedWithoutMissingRemovals;
+				}
+			}
+		}
+	}
+
+	if (
+		matchIndex === -1 &&
+		useFuzzy &&
+		removals.length >= 2 &&
+		contextLines.length === 0
+	) {
+		const firstRemoval = removals[0].content;
+		const lastRemoval = removals[removals.length - 1].content;
+		const firstIdx = findLineIndex(lines, firstRemoval, 0, true);
+		if (firstIdx !== -1) {
+			const rangeEnd = firstIdx + expected.length - 1;
+			if (rangeEnd < lines.length) {
+				const lastInRange = lines[rangeEnd];
+				let lastMatches = lastInRange === lastRemoval;
+				if (!lastMatches) {
+					for (const level of NORMALIZATION_LEVELS.slice(1)) {
+						if (
+							normalizeWhitespace(lastInRange, level) ===
+							normalizeWhitespace(lastRemoval, level)
+						) {
+							lastMatches = true;
+							break;
+						}
+					}
+				}
+				if (lastMatches) {
+					let matchCount = 0;
+					for (let k = 0; k < expected.length; k++) {
+						const fileLine = lines[firstIdx + k];
+						const expLine = expected[k];
+						if (fileLine === expLine) {
+							matchCount++;
+							continue;
+						}
+						for (const level of NORMALIZATION_LEVELS.slice(1)) {
+							if (
+								normalizeWhitespace(fileLine, level) ===
+								normalizeWhitespace(expLine, level)
+							) {
+								matchCount++;
+								break;
+							}
+						}
+					}
+					const matchRatio = matchCount / expected.length;
+					if (matchRatio >= 0.5) {
+						matchIndex = firstIdx;
+						matchedExpected = lines.slice(firstIdx, firstIdx + expected.length);
+					}
+				}
 			}
 		}
 	}
@@ -552,7 +623,7 @@ function applyHunkToLines(
 		throw new Error(errorMsg);
 	}
 
-	const deleteCount = hasExpected ? expected.length : 0;
+	const deleteCount = hasExpected ? matchedExpected.length : 0;
 	const originalIndex = matchIndex;
 	const oldStart = Math.min(
 		originalLines.length,
@@ -561,10 +632,10 @@ function applyHunkToLines(
 	const newStart = matchIndex + 1;
 
 	const adjustedReplacement =
-		useFuzzy && hasExpected
+		useFuzzy && hasExpected && matchedExpected.length === expected.length
 			? adjustReplacementIndentation(
 					hunk,
-					lines.slice(matchIndex, matchIndex + expected.length),
+					lines.slice(matchIndex, matchIndex + matchedExpected.length),
 					originalLines,
 				)
 			: replacement;

package/src/core/src/tools/builtin/patch/parse.ts

@@ -14,7 +14,7 @@ export function parsePatchInput(patch: string): {
 		throw new Error('Patch content is empty.');
 	}
 
-	if (trimmed.includes(PATCH_BEGIN_MARKER)) {
+	if (trimmed.startsWith(PATCH_BEGIN_MARKER)) {
 		return {
 			format: 'enveloped',
 			operations: parseEnvelopedPatch(patch),

package/src/core/src/tools/builtin/patch/repair.ts

@@ -37,7 +37,8 @@ function appendMissingEndMarker(patch: string): string {
 	const hasContent =
 		trimmed.includes('*** Update File:') ||
 		trimmed.includes('*** Add File:') ||
-		trimmed.includes('*** Delete File:');
+		trimmed.includes('*** Delete File:') ||
+		trimmed.includes('*** Replace in:');
 
 	if (hasContent) {
 		return `${trimmed}\n${PATCH_END_MARKER}`;

package/src/core/src/tools/builtin/patch.txt

@@ -1,207 +1,94 @@
-Apply a patch to modify one or more files. The tool accepts the **enveloped patch format** (preferred) and will also auto-convert standard unified diffs (`--- / +++`) if you provide one.
+Apply a patch to modify one or more files.
 
-**Quick checklist before you call the tool:**
-- Finished patch must include both `*** Begin Patch` and `*** End Patch`
-- Each change needs a directive (`*** Add File`, `*** Update File`, `*** Delete File`)
-- Include real context lines (prefixed with space) around your changes
-- Keep paths relative to the project root
-- **Use multiple `@@` hunks for multiple edits to the same file - do NOT make separate tool calls**
+Use the **enveloped format** by default. Standard unified diffs (`---` / `+++`) are also accepted.
 
-**RECOMMENDED: Use apply_patch for targeted file edits to avoid rewriting entire files and wasting tokens.**
+## Fastest / safest mode (recommended): Replace
 
-**FUZZY MATCHING**: By default, fuzzy matching is enabled to handle whitespace differences (tabs vs spaces).
-Exact matching is tried first, then normalized matching if exact fails. Disable with `fuzzyMatch: false` if needed.
+Use this when exact `+/-/context` hunks are hard to build.
 
-Use `apply_patch` only when:
-- You want to make targeted edits to specific lines (primary use case)
-- You want to make multiple related changes across different files in a single operation
-- You need to add/delete entire files along with modifications
-- You have JUST read the file immediately before (within the same response) and are confident the content hasn't changed
-
-**CRITICAL - ALWAYS READ BEFORE PATCHING**: You MUST read the file content immediately before creating a patch.
-Never rely on memory or previous reads. Even with fuzzy matching enabled (tolerates tabs vs spaces),
-If the file content has changed significantly since you last read it, the patch may still fail.
-
-**ALLOW REJECTS**: If you are applying a patch that might include stale hunks, set `allowRejects: true`.
-The tool will apply the hunks it can and skip the rest, returning the skipped hunks with reasons.
-
-**ALREADY APPLIED?**: If the removal lines are already gone or the additions are already present, the tool will treat the hunk as applied and move on—no need to resend the same change.
-
-**Alternative: Use the `edit` tool if you need fuzzy matching or structured operations.**
-
-## ⚠️ CRITICAL: Multiple Edits to Same File
-
-**DO NOT make separate `apply_patch` calls for the same file!**
-
-Instead, use multiple `@@` hunks in a single patch:
-
-```
+```text
 *** Begin Patch
-*** Update File: src/app.ts
-@@ first change - line 10
- function init() {
--  const port = 3000;
-+  const port = 8080;
-   return port;
- }
-@@ second change - line 25
- function start() {
--  console.log("Starting...");
-+  console.log("Server starting...");
-   init();
- }
-@@ third change - line 40
- function cleanup() {
--  // TODO
-+  console.log("Cleaning up...");
- }
+*** Replace in: path/to/file.ts
+*** Find:
+old text block
+*** With:
+new text block
 *** End Patch
 ```
 
-**This makes ONE tool call instead of three, saving tokens and reducing latency.**
-
-## Patch Formats
+You can include multiple `*** Find:` / `*** With:` pairs under one `*** Replace in:` block.
 
-### Preferred: Enveloped Patch
+---
 
-All patches must be wrapped in markers and use explicit file directives:
+## Standard mode: Add / Update / Delete
 
-```
+```text
 *** Begin Patch
-*** Add File: path/to/file.txt
-+line 1
-+line 2
-*** Update File: path/to/other.txt
+*** Add File: path/to/new.txt
++new content
+*** Update File: path/to/existing.ts
+ context line
 -old line
 +new line
-*** Delete File: path/to/delete.txt
+*** Delete File: path/to/old.txt
 *** End Patch
 ```
 
-### Also Supported: Standard Unified Diff
+For multiple edits in the same file, use **one** `*** Update File:` block with multiple `@@` hunks.
 
-You can paste a regular `git diff` style patch:
+---
 
-```
-diff --git a/src/app.ts b/src/app.ts
---- a/src/app.ts
-+++ b/src/app.ts
-@@
--const PORT = 3000;
-+const PORT = 8080;
-```
+## Rules that prevent failures
 
-The tool will convert it automatically, but the enveloped format is still recommended because it is more explicit and easier to control.
+1. Patch must start with `*** Begin Patch` and end with `*** End Patch`.
+2. Paths must be relative to project root.
+3. In `Update File`, `-` lines must match file content exactly.
+4. Include real context lines (prefix with a single space ` `).
+5. `@@` lines are only hints; they do not replace real context.
+6. Do not make separate `apply_patch` calls for the same file in one task.
 
-## File Operations
+---
 
-### Add a new file:
-```
-*** Begin Patch
-*** Add File: src/hello.ts
-+export function hello() {
-+  console.log("Hello!");
-+}
-*** End Patch
-```
+## Matching behavior
 
-### Update an existing file (simple replacement):
-```
-*** Begin Patch
-*** Update File: src/config.ts
--const PORT = 3000;
-+const PORT = 8080;
-*** End Patch
-```
+- `fuzzyMatch` defaults to `true` (helps with small whitespace differences).
+- Set `fuzzyMatch: false` for strict matching.
+- If a hunk may be stale, set `allowRejects: true` to apply valid parts and skip invalid ones.
 
-**CRITICAL**: The `-` lines must match EXACTLY what's in the file, character-for-character. If you're not 100% certain, use the `edit` tool instead.
+---
 
-### Update with context (recommended for precision):
-```
-*** Begin Patch
-*** Update File: src/app.ts
-@@ function main()
- function main() {
--  console.log("old");
-+  console.log("new");
- }
-*** End Patch
-```
+## Common errors
 
-**IMPORTANT**: 
-- The `@@` line is an OPTIONAL hint to help locate the change - it's a comment, not parsed as context
-- REQUIRED: Actual context lines (starting with space ` `) that match the file exactly
-- The context lines with space prefix are what the tool uses to find the location
-- The `@@` line just helps humans/AI understand what section you're editing
+- **Missing end marker** → add `*** End Patch`.
+- **Failed to find expected lines** → re-read the file and rebuild patch with exact context.
+- **Ambiguous update** → provide longer context or use `*** Replace in:` mode.
 
+---
 
-### Update multiple locations in the same file:
-```
+## Minimal templates
+
+### Update one line
+```text
 *** Begin Patch
 *** Update File: src/app.ts
-@@ first section - near line 10
- function init() {
--  const port = 3000;
-+  const port = 8080;
-   return port;
- }
-@@ second section - near line 25
- function start() {
--  console.log("Starting...");
-+  console.log("Server starting...");
-   init();
+ function main() {
+-  console.log("old");
++  console.log("new");
  }
 *** End Patch
 ```
 
-**IMPORTANT**: Use separate `@@` headers for each non-consecutive change location. This allows multiple edits to the same file in one patch, saving tokens and reducing tool calls.
-
-### Delete a file:
-```
+### Add file
+```text
 *** Begin Patch
-*** Delete File: old/unused.ts
+*** Add File: src/new.ts
++export const ok = true;
 *** End Patch
 ```
 
-### Multiple operations in one patch:
-```
+### Delete file
+```text
 *** Begin Patch
-*** Add File: new.txt
-+New content
-*** Update File: existing.txt
--old
-+new
-*** Delete File: obsolete.txt
+*** Delete File: src/obsolete.ts
 *** End Patch
 ```
-
-## Line Prefixes
-
-- Lines starting with `+` are added
-- Lines starting with `-` are removed  
-- Lines starting with ` ` (space) are context (kept unchanged)
-- Lines starting with `@@` are optional hints/comments (not parsed as context)
-
-## Common Errors
-
-**"Missing *** End Patch marker"**: Every patch MUST end with `*** End Patch` on its own line.
-This is the most common error. Double-check your patch ends with:
-```
-*** End Patch
-```
-
-**"Failed to find expected lines"**: The file content doesn't match your patch. Common causes:
-- Missing context lines (lines with space prefix)
-- Using `@@` line as context instead of real context lines
-- The file content has changed since you read it
-- Whitespace/indentation mismatch
-
-**Solution**: Always read the file immediately before patching and include actual context lines with space prefix.
-
-## Important Notes
-
-- **Patches are fragile**: Any mismatch in whitespace, indentation, or content will cause failure
-- **Use `edit` for reliability**: The `edit` tool can make targeted changes without requiring exact matches
-- All file paths are relative to the project root
-- The enveloped format is the most reliable; unified diffs are converted automatically if provided
-- Always wrap patches with `*** Begin Patch` and `*** End Patch` when you write them manually

package/src/index.ts

@@ -70,6 +70,8 @@ export {
 export type {
 	SetuAuth,
 	SetuProviderOptions,
+	SetuPaymentCallbacks,
+	SetuBalanceUpdate,
 	SetuBalanceResponse,
 	SolanaUsdcBalanceResponse,
 } from './providers/src/index.ts';

package/src/providers/src/catalog-manual.ts

@@ -30,6 +30,12 @@ const isAllowedGoogleModel = (id: string): boolean => {
 	return false;
 };
 
+const isAllowedZaiModel = (id: string): boolean => {
+	if (id.startsWith('glm-4.7')) return true;
+	if (id.startsWith('glm-5')) return true;
+	return false;
+};
+
 const SETU_SOURCES: Array<{
 	id: ProviderId;
 	npm: string;
@@ -55,6 +61,11 @@ const SETU_SOURCES: Array<{
 		npm: '@ai-sdk/google',
 		family: 'google',
 	},
+	{
+		id: 'zai',
+		npm: '@ai-sdk/openai-compatible',
+		family: 'openai-compatible',
+	},
 ];
 
 function cloneModel(model: ModelInfo): ModelInfo {
@@ -83,6 +94,7 @@ function buildSetuEntry(base: CatalogMap): ProviderCatalogEntry | null {
 			if (id === 'openai') return isAllowedOpenAIModel(model.id);
 			if (id === 'anthropic') return isAllowedAnthropicModel(model.id);
 			if (id === 'google') return isAllowedGoogleModel(model.id);
+			if (id === 'zai') return isAllowedZaiModel(model.id);
 			return true;
 		});
 		return sourceModels.map((model) => {

package/src/providers/src/catalog.ts

@@ -703,6 +703,31 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 					output: 128000,
 				},
 			},
+			{
+				id: 'gpt-5.3-codex-spark',
+				label: 'GPT-5.3 Codex Spark',
+				modalities: {
+					input: ['text', 'image', 'pdf'],
+					output: ['text'],
+				},
+				toolCall: true,
+				reasoningText: true,
+				attachment: true,
+				temperature: false,
+				knowledge: '2025-08-31',
+				releaseDate: '2026-02-05',
+				lastUpdated: '2026-02-05',
+				openWeights: false,
+				cost: {
+					input: 1.75,
+					output: 14,
+					cacheRead: 0.175,
+				},
+				limit: {
+					context: 128000,
+					output: 32000,
+				},
+			},
 			{
 				id: 'o1',
 				label: 'o1',
@@ -5229,29 +5254,6 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 					output: 100000,
 				},
 			},
-			{
-				id: 'openrouter/pony-alpha',
-				label: 'Pony Alpha',
-				modalities: {
-					input: ['text'],
-					output: ['text'],
-				},
-				toolCall: true,
-				reasoningText: true,
-				attachment: false,
-				temperature: true,
-				releaseDate: '2026-02-06',
-				lastUpdated: '2026-02-06',
-				openWeights: false,
-				cost: {
-					input: 0,
-					output: 0,
-				},
-				limit: {
-					context: 200000,
-					output: 131000,
-				},
-			},
 			{
 				id: 'openrouter/sherlock-dash-alpha',
 				label: 'Sherlock Dash Alpha',
@@ -6043,6 +6045,55 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 					output: 8192,
 				},
 			},
+			{
+				id: 'stepfun/step-3.5-flash',
+				label: 'Step 3.5 Flash',
+				modalities: {
+					input: ['text'],
+					output: ['text'],
+				},
+				toolCall: true,
+				reasoningText: true,
+				attachment: false,
+				temperature: true,
+				knowledge: '2025-01',
+				releaseDate: '2026-01-29',
+				lastUpdated: '2026-01-29',
+				openWeights: true,
+				cost: {
+					input: 0.1,
+					output: 0.3,
+					cacheRead: 0.02,
+				},
+				limit: {
+					context: 256000,
+					output: 256000,
+				},
+			},
+			{
+				id: 'stepfun/step-3.5-flash:free',
+				label: 'Step 3.5 Flash (free)',
+				modalities: {
+					input: ['text'],
+					output: ['text'],
+				},
+				toolCall: true,
+				reasoningText: true,
+				attachment: false,
+				temperature: true,
+				knowledge: '2025-01',
+				releaseDate: '2026-01-29',
+				lastUpdated: '2026-01-29',
+				openWeights: true,
+				cost: {
+					input: 0,
+					output: 0,
+				},
+				limit: {
+					context: 256000,
+					output: 256000,
+				},
+			},
 			{
 				id: 'thudm/glm-z1-32b:free',
 				label: 'GLM Z1 32B (free)',
@@ -6541,6 +6592,30 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 					output: 65535,
 				},
 			},
+			{
+				id: 'z-ai/glm-5',
+				label: 'GLM-5',
+				modalities: {
+					input: ['text'],
+					output: ['text'],
+				},
+				toolCall: true,
+				reasoningText: true,
+				attachment: false,
+				temperature: true,
+				releaseDate: '2026-02-12',
+				lastUpdated: '2026-02-12',
+				openWeights: true,
+				cost: {
+					input: 1,
+					output: 3.2,
+					cacheRead: 0.2,
+				},
+				limit: {
+					context: 202752,
+					output: 131000,
+				},
+			},
 		],
 		label: 'OpenRouter',
 		env: ['OPENROUTER_API_KEY'],
@@ -7341,6 +7416,31 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 					npm: '@ai-sdk/anthropic',
 				},
 			},
+			{
+				id: 'minimax-m2.5-free',
+				label: 'MiniMax M2.5 Free',
+				modalities: {
+					input: ['text'],
+					output: ['text'],
+				},
+				toolCall: true,
+				reasoningText: true,
+				attachment: false,
+				temperature: true,
+				knowledge: '2025-01',
+				releaseDate: '2026-02-12',
+				lastUpdated: '2026-02-12',
+				openWeights: true,
+				cost: {
+					input: 0,
+					output: 0,
+					cacheRead: 0,
+				},
+				limit: {
+					context: 204800,
+					output: 131072,
+				},
+			},
 			{
 				id: 'qwen3-coder',
 				label: 'Qwen3 Coder',
@@ -7603,6 +7703,31 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 					output: 131072,
 				},
 			},
+			{
+				id: 'glm-5',
+				label: 'GLM-5',
+				modalities: {
+					input: ['text'],
+					output: ['text'],
+				},
+				toolCall: true,
+				reasoningText: true,
+				attachment: false,
+				temperature: true,
+				releaseDate: '2026-02-11',
+				lastUpdated: '2026-02-11',
+				openWeights: false,
+				cost: {
+					input: 1,
+					output: 3.2,
+					cacheRead: 0.2,
+					cacheWrite: 0,
+				},
+				limit: {
+					context: 204800,
+					output: 131072,
+				},
+			},
 		],
 		label: 'Z.AI',
 		env: ['ZHIPU_API_KEY'],
@@ -7817,6 +7942,31 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 					output: 131072,
 				},
 			},
+			{
+				id: 'glm-5',
+				label: 'GLM-5',
+				modalities: {
+					input: ['text'],
+					output: ['text'],
+				},
+				toolCall: true,
+				reasoningText: true,
+				attachment: false,
+				temperature: true,
+				releaseDate: '2026-02-11',
+				lastUpdated: '2026-02-11',
+				openWeights: false,
+				cost: {
+					input: 0,
+					output: 0,
+					cacheRead: 0,
+					cacheWrite: 0,
+				},
+				limit: {
+					context: 204800,
+					output: 131072,
+				},
+			},
 		],
 		label: 'Z.AI Coding Plan',
 		env: ['ZHIPU_API_KEY'],
@@ -8008,7 +8158,7 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 				},
 				limit: {
 					context: 128000,
-					output: 16000,
+					output: 32000,
 				},
 			},
 			{
@@ -8032,7 +8182,7 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 				},
 				limit: {
 					context: 128000,
-					output: 16000,
+					output: 32000,
 				},
 			},
 			{
@@ -8128,7 +8278,7 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 				},
 				limit: {
 					context: 128000,
-					output: 16000,
+					output: 32000,
 				},
 			},
 			{
@@ -8223,7 +8373,7 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 					output: 0,
 				},
 				limit: {
-					context: 128000,
+					context: 64000,
 					output: 16384,
 				},
 			},
@@ -8320,7 +8470,7 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 				},
 				limit: {
 					context: 128000,
-					output: 128000,
+					output: 64000,
 				},
 			},
 			{
@@ -8392,7 +8542,7 @@ export const catalog: Partial<Record<ProviderId, ProviderCatalogEntry>> = {
 				},
 				limit: {
 					context: 128000,
-					output: 100000,
+					output: 128000,
 				},
 			},
 			{

package/src/providers/src/index.ts

@@ -35,6 +35,7 @@ export type {
 	SetuAuth,
 	SetuProviderOptions,
 	SetuPaymentCallbacks,
+	SetuBalanceUpdate,
 	SetuBalanceResponse,
 	SolanaUsdcBalanceResponse,
 } from './setu-client.ts';

package/src/providers/src/setu-client.ts

@@ -44,6 +44,13 @@ const DEFAULT_RPC_URL = 'https://api.mainnet-beta.solana.com';
 const DEFAULT_MAX_ATTEMPTS = 3;
 const DEFAULT_MAX_PAYMENT_ATTEMPTS = 20;
 
+export type SetuBalanceUpdate = {
+	costUsd: number;
+	balanceRemaining: number;
+	inputTokens?: number;
+	outputTokens?: number;
+};
+
 export type SetuPaymentCallbacks = {
 	onPaymentRequired?: (amountUsd: number, currentBalance?: number) => void;
 	onPaymentSigning?: () => void;
@@ -57,6 +64,7 @@ export type SetuPaymentCallbacks = {
 		amountUsd: number;
 		currentBalance: number;
 	}) => Promise<'crypto' | 'fiat' | 'cancel'>;
+	onBalanceUpdate?: (update: SetuBalanceUpdate) => void;
 };
 
 export type SetuProviderOptions = {
@@ -197,7 +205,7 @@ export function createSetuFetch(
 			const response = await baseFetch(input, { ...init, body, headers });
 
 			if (response.status !== 402) {
-				return response;
+				return wrapResponseWithBalanceSniffing(response, callbacks);
 			}
 
 			const payload = await response.json().catch(() => ({}));
@@ -266,6 +274,70 @@ export function createSetuFetch(
 	};
 }
 
+function tryParseSetuComment(
+	line: string,
+	onBalanceUpdate: (update: SetuBalanceUpdate) => void,
+) {
+	const trimmed = line.replace(/\r$/, '');
+	if (!trimmed.startsWith(': setu ')) return;
+	try {
+		const data = JSON.parse(trimmed.slice(7));
+		onBalanceUpdate({
+			costUsd: parseFloat(data.cost_usd ?? '0'),
+			balanceRemaining: parseFloat(data.balance_remaining ?? '0'),
+			inputTokens: data.input_tokens ? Number(data.input_tokens) : undefined,
+			outputTokens: data.output_tokens ? Number(data.output_tokens) : undefined,
+		});
+	} catch {}
+}
+
+function wrapResponseWithBalanceSniffing(
+	response: Response,
+	callbacks: SetuPaymentCallbacks,
+): Response {
+	if (!callbacks.onBalanceUpdate) return response;
+
+	const balanceHeader = response.headers.get('x-balance-remaining');
+	const costHeader = response.headers.get('x-cost-usd');
+	if (balanceHeader && costHeader) {
+		callbacks.onBalanceUpdate({
+			costUsd: parseFloat(costHeader),
+			balanceRemaining: parseFloat(balanceHeader),
+		});
+		return response;
+	}
+
+	if (!response.body) return response;
+
+	const onBalanceUpdate = callbacks.onBalanceUpdate;
+	let partial = '';
+	const decoder = new TextDecoder();
+	const transform = new TransformStream<Uint8Array, Uint8Array>({
+		transform(chunk, controller) {
+			controller.enqueue(chunk);
+			partial += decoder.decode(chunk, { stream: true });
+			let nlIndex = partial.indexOf('\n');
+			while (nlIndex !== -1) {
+				const line = partial.slice(0, nlIndex);
+				partial = partial.slice(nlIndex + 1);
+				tryParseSetuComment(line, onBalanceUpdate);
+				nlIndex = partial.indexOf('\n');
+			}
+		},
+		flush() {
+			if (partial.trim()) {
+				tryParseSetuComment(partial, onBalanceUpdate);
+			}
+		},
+	});
+
+	return new Response(response.body.pipeThrough(transform), {
+		status: response.status,
+		statusText: response.statusText,
+		headers: response.headers,
+	});
+}
+
 /**
  * Create a Setu-backed AI model.
  *