Muse Spark 1.1

Model ID

Use with the Workers AI binding, the AI Gateway REST API, or the Unified API.

Calling the provider directly? Use the provider's model name instead, including with provider-native endpoints or the provider's own SDK. For example, Anthropic's name for anthropic/claude-opus-4.8 is claude-opus-4-8.

Third-party

Muse Spark 1.1 is Meta's agentic and coding model for multi-step tool loops, software engineering assistants, and long-context reasoning.

Model Info
More information	link ↗
Request formats	Responses
Pricing	Input (per 1M tokens)$1.25 Cached input (per 1M tokens)$0.16 Output (per 1M tokens)$4.25

Usage

const response = await env.AI.run(
  'meta/muse-spark-1.1',
  {
    messages: [
      {
        role: 'system',
        content:
          'You are a pragmatic senior JavaScript reviewer. Prefer small, production-safe fixes and explain the root cause before the patch.',
      },
      {
        role: 'user',
        content:
          'Review this function:\n\nasync function processOnce(queue) {\n  if (queue.length === 0) return;\n  const item = queue.shift();\n  await save(item);\n  if (queue.length > 0) processOnce(queue);\n}\n\nTwo requests can call processOnce(queue) at the same time. What can go wrong, and how would you fix it without changing the save() function? Include a concise corrected implementation.',
      },
    ],
    max_completion_tokens: 4096,
  },
)
console.log(response)

curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/run \
  --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "meta/muse-spark-1.1",
  "input": {
    "messages": [
      {
        "role": "system",
        "content": "You are a pragmatic senior JavaScript reviewer. Prefer small, production-safe fixes and explain the root cause before the patch."
      },
      {
        "role": "user",
        "content": "Review this function:\n\nasync function processOnce(queue) {\n  if (queue.length === 0) return;\n  const item = queue.shift();\n  await save(item);\n  if (queue.length > 0) processOnce(queue);\n}\n\nTwo requests can call processOnce(queue) at the same time. What can go wrong, and how would you fix it without changing the save() function? Include a concise corrected implementation."
      }
    ],
    "max_completion_tokens": 4096
  }
}'

This is a concurrent drain + fire-and-forget bug.

### Root cause

JS is single-threaded, but `await` yields to the event loop. You have shared mutable state `queue` with no mutual exclusion.

If A and B call `processOnce(queue)` at the same time:

1.  A: `queue.length > 0` -> true
2.  A: `shift()` -> item1, `await save(item1)` -> **yields**
3.  B: `queue.length > 0` -> still true, `shift()` -> item2, `await save(item2)` -> **now `save()` is running in parallel**. If `save()` must be serial / order dependent, you break it. `shift()` can also return `undefined` if B empties it first.
4.  `if(queue.length > 0) processOnce(queue)` is **not awaited**. Caller resolves early, queue is still draining. Rejections are lost and caller thinks it's done.

### Fix

Serialize draining. One drain loop at a time, `while` not recursion, and piggyback concurrent callers onto the same promise + re-check for items enqueued in the gap.

Don't change `save()`. Just guard it:

```js
let drainPromise = null;

async function processOnce(queue) {
  if (drainPromise) return drainPromise;

  drainPromise = (async () => {
    while (queue.length > 0) {
      const item = queue.shift();
      await save(item);
    }
  })().finally(() => {
    drainPromise = null;
    // items pushed between last shift() and clearing the flag
    if (queue.length > 0) return processOnce(queue);
  });

  return drainPromise;
}
```

What this changes:

* `drainPromise` lock prevents 2 concurrent drain loops.
* `while` loop + `await save()` guarantees serial saves.
* Concurrent callers return the same promise, so they properly await completion.
* `finally` re-check prevents losing items enqueued right after the loop ends.

{
  "id": "resp_6ac7afe7b1edff27f7904b24",
  "object": "response",
  "created_at": 1791471590,
  "model": "muse-spark-1.1",
  "output": [
    {
      "type": "reasoning",
      "id": "rs_6ac7afe7b1edff27f7904b24:rs_d69dde63-919d-49f4-9c17-65b594289afb",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "message",
      "id": "msg_ce37fe35-31f4-4a9c-aea7-3f2ee96186e5",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "This is a concurrent drain + fire-and-forget bug.\n\n### Root cause\n\nJS is single-threaded, but `await` yields to the event loop. You have shared mutable state `queue` with no mutual exclusion.\n\nIf A and B call `processOnce(queue)` at the same time:\n\n1.  A: `queue.length > 0` -> true\n2.  A: `shift()` -> item1, `await save(item1)` -> **yields**\n3.  B: `queue.length > 0` -> still true, `shift()` -> item2, `await save(item2)` -> **now `save()` is running in parallel**. If `save()` must be serial / order dependent, you break it. `shift()` can also return `undefined` if B empties it first.\n4.  `if(queue.length > 0) processOnce(queue)` is **not awaited**. Caller resolves early, queue is still draining. Rejections are lost and caller thinks it's done.\n\n### Fix\n\nSerialize draining. One drain loop at a time, `while` not recursion, and piggyback concurrent callers onto the same promise + re-check for items enqueued in the gap.\n\nDon't change `save()`. Just guard it:\n\n```js\nlet drainPromise = null;\n\nasync function processOnce(queue) {\n  if (drainPromise) return drainPromise;\n\n  drainPromise = (async () => {\n    while (queue.length > 0) {\n      const item = queue.shift();\n      await save(item);\n    }\n  })().finally(() => {\n    drainPromise = null;\n    // items pushed between last shift() and clearing the flag\n    if (queue.length > 0) return processOnce(queue);\n  });\n\n  return drainPromise;\n}\n```\n\nWhat this changes:\n\n* `drainPromise` lock prevents 2 concurrent drain loops.\n* `while` loop + `await save()` guarantees serial saves.\n* Concurrent callers return the same promise, so they properly await completion.\n* `finally` re-check prevents losing items enqueued right after the loop ends.",
          "annotations": [],
          "logprobs": []
        }
      ],
      "status": "completed"
    }
  ],
  "status": "completed",
  "usage": {
    "input_tokens": 134,
    "output_tokens": 2852,
    "total_tokens": 2986,
    "input_tokens_details": {
      "cached_tokens": 113
    },
    "output_tokens_details": {
      "reasoning_tokens": 2409
    }
  },
  "top_logprobs": 0,
  "temperature": 1,
  "top_p": 1,
  "service_tier": "auto",
  "reasoning": {
    "effort": "high"
  },
  "background": false,
  "tools": [],
  "tool_choice": "auto",
  "truncation": "disabled",
  "completed_at": 1791471591,
  "error": null,
  "incomplete_details": null,
  "instructions": null,
  "parallel_tool_calls": true,
  "store": true,
  "max_output_tokens": 4096
}

Examples

SQL Correctness Review — Review a SQL query for correctness and performance, then propose a safer replacement.

const response = await env.AI.run(
  'meta/muse-spark-1.1',
  {
    messages: [
      {
        role: 'system',
        content:
          'You are a database engineer. Be precise about semantics, indexes, and concurrency. Do not assume schema details that are not provided.',
      },
      {
        role: 'user',
        content:
          "A payments table has columns id, account_id, status, amount_cents, created_at, and settled_at. Review this query used to calculate an account's settled balance: SELECT SUM(amount_cents) FROM payments WHERE account_id = ? AND status = 'settled' OR status = 'refunded' AND created_at >= ?; Identify the correctness bug, give the corrected query, and suggest one useful index. Keep the answer under 400 words.",
      },
    ],
    max_completion_tokens: 4096,
  },
)
console.log(response)

curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/run \
  --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "meta/muse-spark-1.1",
  "input": {
    "messages": [
      {
        "role": "system",
        "content": "You are a database engineer. Be precise about semantics, indexes, and concurrency. Do not assume schema details that are not provided."
      },
      {
        "role": "user",
        "content": "A payments table has columns id, account_id, status, amount_cents, created_at, and settled_at. Review this query used to calculate an account'\''s settled balance: SELECT SUM(amount_cents) FROM payments WHERE account_id = ? AND status = '\''settled'\'' OR status = '\''refunded'\'' AND created_at >= ?; Identify the correctness bug, give the corrected query, and suggest one useful index. Keep the answer under 400 words."
      }
    ],
    "max_completion_tokens": 4096
  }
}'

Operator precedence bug. `AND` binds tighter than `OR`, so your query is parsed as:

```sql
(account_id = ? AND status = 'settled')
OR 
(status = 'refunded' AND created_at >= ?)
```

Failures:
1.  Leaks data across accounts - all `refunded` rows for any account after `?` are summed into this account.
2.  Drops the `created_at` filter for `settled` rows, and drops `account_id` filter for `refunded`.

### Corrected query

If you intend to filter both statuses by date:

```sql
SELECT COALESCE(SUM(amount_cents),0) FROM payments
WHERE account_id = ?
  AND status IN ('settled','refunded')
  AND created_at >= ?;
```

If you intended date filter only for refunded:

```sql
SELECT COALESCE(SUM(amount_cents),0) FROM payments
WHERE account_id = ?
  AND (status = 'settled' OR (status = 'refunded' AND created_at >= ?));
```

Note: For a *settled* balance, you likely want `settled_at >= ?`, not `created_at`.

### Index

Composite to support equality on `account_id`, equality/IN on `status`, range on `created_at`, covering `amount_cents`:

```sql
CREATE INDEX idx_payments_acct_status_created 
ON payments (account_id, status, created_at) INCLUDE (amount_cents);
```

This allows an index-only scan per account without touching the heap.

{
  "id": "resp_6ac7aff0b2536166015f4c31",
  "object": "response",
  "created_at": 1791471600,
  "model": "muse-spark-1.1",
  "output": [
    {
      "type": "reasoning",
      "id": "rs_6ac7aff0b2536166015f4c31:rs_bd77c866-9452-4cf2-84d6-3052aa3f613f",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "message",
      "id": "msg_58de9d22-820f-49a1-aecd-ff8c43ae16a3",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Operator precedence bug. `AND` binds tighter than `OR`, so your query is parsed as:\n\n```sql\n(account_id = ? AND status = 'settled')\nOR \n(status = 'refunded' AND created_at >= ?)\n```\n\nFailures:\n1.  Leaks data across accounts - all `refunded` rows for any account after `?` are summed into this account.\n2.  Drops the `created_at` filter for `settled` rows, and drops `account_id` filter for `refunded`.\n\n### Corrected query\n\nIf you intend to filter both statuses by date:\n\n```sql\nSELECT COALESCE(SUM(amount_cents),0) FROM payments\nWHERE account_id = ?\n  AND status IN ('settled','refunded')\n  AND created_at >= ?;\n```\n\nIf you intended date filter only for refunded:\n\n```sql\nSELECT COALESCE(SUM(amount_cents),0) FROM payments\nWHERE account_id = ?\n  AND (status = 'settled' OR (status = 'refunded' AND created_at >= ?));\n```\n\nNote: For a *settled* balance, you likely want `settled_at >= ?`, not `created_at`.\n\n### Index\n\nComposite to support equality on `account_id`, equality/IN on `status`, range on `created_at`, covering `amount_cents`:\n\n```sql\nCREATE INDEX idx_payments_acct_status_created \nON payments (account_id, status, created_at) INCLUDE (amount_cents);\n```\n\nThis allows an index-only scan per account without touching the heap.",
          "annotations": [],
          "logprobs": []
        }
      ],
      "status": "completed"
    }
  ],
  "status": "completed",
  "usage": {
    "input_tokens": 143,
    "output_tokens": 1578,
    "total_tokens": 1721,
    "input_tokens_details": {
      "cached_tokens": 113
    },
    "output_tokens_details": {
      "reasoning_tokens": 1249
    }
  },
  "top_logprobs": 0,
  "temperature": 1,
  "top_p": 1,
  "service_tier": "auto",
  "reasoning": {
    "effort": "high"
  },
  "background": false,
  "tools": [],
  "tool_choice": "auto",
  "truncation": "disabled",
  "completed_at": 1791471600,
  "error": null,
  "incomplete_details": null,
  "instructions": null,
  "parallel_tool_calls": true,
  "store": true,
  "max_output_tokens": 4096
}

TypeScript Utility Refactor — Turn an unreliable JavaScript helper into a typed, testable implementation with explicit edge-case behavior.

const response = await env.AI.run(
  'meta/muse-spark-1.1',
  {
    messages: [
      {
        role: 'system',
        content:
          'You are a TypeScript library maintainer. Prefer small APIs, explicit types, and tests that document behavior.',
      },
      {
        role: 'user',
        content:
          'Refactor this helper into TypeScript:\n\nfunction groupBy(items, key) {\n  return items.reduce((out, item) => {\n    (out[item[key]] ||= []).push(item);\n    return out;\n  }, {});\n}\n\nProvide a generic type-safe version and three focused tests covering an empty input, a missing key value, and two items sharing a key. State your chosen behavior for missing keys.',
      },
    ],
    max_completion_tokens: 4096,
  },
)
console.log(response)

curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/run \
  --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "meta/muse-spark-1.1",
  "input": {
    "messages": [
      {
        "role": "system",
        "content": "You are a TypeScript library maintainer. Prefer small APIs, explicit types, and tests that document behavior."
      },
      {
        "role": "user",
        "content": "Refactor this helper into TypeScript:\n\nfunction groupBy(items, key) {\n  return items.reduce((out, item) => {\n    (out[item[key]] ||= []).push(item);\n    return out;\n  }, {});\n}\n\nProvide a generic type-safe version and three focused tests covering an empty input, a missing key value, and two items sharing a key. State your chosen behavior for missing keys."
      }
    ],
    "max_completion_tokens": 4096
  }
}'

Here's a small, strict version. It fixes the original bug where `item[key] === undefined` silently created a `"undefined"` bucket.

### Generic implementation

**Chosen behavior for missing keys:** If `item[key]` is `null` or `undefined`, the item is **skipped** and not included in any group. Rationale: avoids polluting the result with a coerced `"undefined"` key. If you need to keep them, normalize before calling.

```ts
// groupBy.ts
type PropertyKey = string | number | symbol;

// Only allow keys whose value can be a property key (or optional version)
type GroupableKeys<T> = {
  [K in keyof T]: T[K] extends PropertyKey | null | undefined ? K : never
}[keyof T];

export function groupBy<T, K extends GroupableKeys<T>>(
  items: readonly T[],
  key: K
): Record<string, T[]> {
  const out: Record<string, T[]> = {};

  for (const item of items) {
    const raw = item[key];
    if (raw == null) continue; // <-- explicit missing-key policy

    const groupKey = String(raw as PropertyKey);
    (out[groupKey] ??= []).push(item);
  }

  return out;
}
```

> Typed to `Record<string, T[]>` intentionally. If you want literal key inference you can change the return to `Record<Extract<T[K], PropertyKey> & {}, T[]>` but for a library this is more ergonomic and doesn't lie about runtime `String()` coercion.

### Tests that document behavior

```ts
// groupBy.test.ts
import { describe, it, expect } from 'vitest';
import { groupBy } from './groupBy';

type Item = { cat?: string; id: number };

describe('groupBy', () => {
  it('empty input -> empty object', () => {
    expect(groupBy([] as Item[], 'cat')).toEqual({});
  });

  it('missing key value -> item skipped', () => {
    const items: Item[] = [
      { cat: 'a', id: 1 },
      { id: 2 }, // cat is missing
      { cat: undefined, id: 3 }, // cat is explicitly undefined
    ];

    const result = groupBy(items, 'cat');

    // only 'a' group exists, no "undefined" bucket
    expect(result).toEqual({
      a: [{ cat: 'a', id: 1 }],
    });
    expect(Object.keys(result)).not.toContain('undefined');
  });

  it('two items sharing a key -> grouped together', () => {
    const items: Item[] = [
      { cat: 'x', id: 1 },
      { cat: 'x', id: 2 },
      { cat: 'y', id: 3 },
    ];

    const result = groupBy(items, 'cat');

    expect(result).toEqual({
      x: [
        { cat: 'x', id: 1 },
        { cat: 'x', id: 2 },
      ],
      y: [{ cat: 'y', id: 3 }],
    });
  });
});
```

{
  "id": "resp_6ac7b0009eadac3fd3fd4b77",
  "object": "response",
  "created_at": 1791471616,
  "model": "muse-spark-1.1",
  "output": [
    {
      "type": "reasoning",
      "id": "rs_6ac7b0009eadac3fd3fd4b77:rs_70691cf1-3099-4eaf-8a65-6c9811ee8913",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "message",
      "id": "msg_20968e1e-1419-4226-89cb-2c6497d11b9e",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Here's a small, strict version. It fixes the original bug where `item[key] === undefined` silently created a `\"undefined\"` bucket.\n\n### Generic implementation\n\n**Chosen behavior for missing keys:** If `item[key]` is `null` or `undefined`, the item is **skipped** and not included in any group. Rationale: avoids polluting the result with a coerced `\"undefined\"` key. If you need to keep them, normalize before calling.\n\n```ts\n// groupBy.ts\ntype PropertyKey = string | number | symbol;\n\n// Only allow keys whose value can be a property key (or optional version)\ntype GroupableKeys<T> = {\n  [K in keyof T]: T[K] extends PropertyKey | null | undefined ? K : never\n}[keyof T];\n\nexport function groupBy<T, K extends GroupableKeys<T>>(\n  items: readonly T[],\n  key: K\n): Record<string, T[]> {\n  const out: Record<string, T[]> = {};\n\n  for (const item of items) {\n    const raw = item[key];\n    if (raw == null) continue; // <-- explicit missing-key policy\n\n    const groupKey = String(raw as PropertyKey);\n    (out[groupKey] ??= []).push(item);\n  }\n\n  return out;\n}\n```\n\n> Typed to `Record<string, T[]>` intentionally. If you want literal key inference you can change the return to `Record<Extract<T[K], PropertyKey> & {}, T[]>` but for a library this is more ergonomic and doesn't lie about runtime `String()` coercion.\n\n### Tests that document behavior\n\n```ts\n// groupBy.test.ts\nimport { describe, it, expect } from 'vitest';\nimport { groupBy } from './groupBy';\n\ntype Item = { cat?: string; id: number };\n\ndescribe('groupBy', () => {\n  it('empty input -> empty object', () => {\n    expect(groupBy([] as Item[], 'cat')).toEqual({});\n  });\n\n  it('missing key value -> item skipped', () => {\n    const items: Item[] = [\n      { cat: 'a', id: 1 },\n      { id: 2 }, // cat is missing\n      { cat: undefined, id: 3 }, // cat is explicitly undefined\n    ];\n\n    const result = groupBy(items, 'cat');\n\n    // only 'a' group exists, no \"undefined\" bucket\n    expect(result).toEqual({\n      a: [{ cat: 'a', id: 1 }],\n    });\n    expect(Object.keys(result)).not.toContain('undefined');\n  });\n\n  it('two items sharing a key -> grouped together', () => {\n    const items: Item[] = [\n      { cat: 'x', id: 1 },\n      { cat: 'x', id: 2 },\n      { cat: 'y', id: 3 },\n    ];\n\n    const result = groupBy(items, 'cat');\n\n    expect(result).toEqual({\n      x: [\n        { cat: 'x', id: 1 },\n        { cat: 'x', id: 2 },\n      ],\n      y: [{ cat: 'y', id: 3 }],\n    });\n  });\n});\n```",
          "annotations": [],
          "logprobs": []
        }
      ],
      "status": "completed"
    }
  ],
  "status": "completed",
  "usage": {
    "input_tokens": 127,
    "output_tokens": 3158,
    "total_tokens": 3285,
    "input_tokens_details": {
      "cached_tokens": 0
    },
    "output_tokens_details": {
      "reasoning_tokens": 2479
    }
  },
  "top_logprobs": 0,
  "temperature": 1,
  "top_p": 1,
  "service_tier": "auto",
  "reasoning": {
    "effort": "high"
  },
  "background": false,
  "tools": [],
  "tool_choice": "auto",
  "truncation": "disabled",
  "completed_at": 1791471616,
  "error": null,
  "incomplete_details": null,
  "instructions": null,
  "parallel_tool_calls": true,
  "store": true,
  "max_output_tokens": 4096
}

Production Incident Triage — Triage a production incident from sparse symptoms and produce an ordered investigation plan.

const response = await env.AI.run(
  'meta/muse-spark-1.1',
  {
    messages: [
      {
        role: 'system',
        content:
          'You are an incident commander. Separate facts from hypotheses, prioritize customer impact, and give reversible next steps before risky remediation.',
      },
      {
        role: 'user',
        content:
          'At 14:05 UTC, API p95 latency rose from 180 ms to 2.4 s for one region. Error rate is 0.8%, database CPU is 45%, cache hit rate fell from 92% to 61%, and deploys have been quiet for six hours. A single endpoint is responsible for 70% of the extra latency. Write an incident update, list the three highest-value checks in order, and propose a safe mitigation.',
      },
    ],
    max_completion_tokens: 4096,
  },
)
console.log(response)

curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/run \
  --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "meta/muse-spark-1.1",
  "input": {
    "messages": [
      {
        "role": "system",
        "content": "You are an incident commander. Separate facts from hypotheses, prioritize customer impact, and give reversible next steps before risky remediation."
      },
      {
        "role": "user",
        "content": "At 14:05 UTC, API p95 latency rose from 180 ms to 2.4 s for one region. Error rate is 0.8%, database CPU is 45%, cache hit rate fell from 92% to 61%, and deploys have been quiet for six hours. A single endpoint is responsible for 70% of the extra latency. Write an incident update, list the three highest-value checks in order, and propose a safe mitigation."
      }
    ],
    "max_completion_tokens": 4096
  }
}'

### INCIDENT UPDATE #1 - 14:20 UTC - API Latency Elevation [Single Region]

**Status:** Investigating - SEV2
**Impact:** Customers served from `us-east-1 [example]` experiencing API p95 of 2.4s vs normal 180ms since 14:05 UTC. p50 stable. Error rate still low at 0.8% but we are seeing timeouts. Other regions healthy.
**Scope:** ~70% of added latency is isolated to a single endpoint: `GET /v1/<hot-endpoint>`

**FACTS:**
* 14:05 UTC start, sustained. Single region only.
* p95: 180ms -> 2.4s. Error rate: 0.8%
* Cache hit rate: 92% -> 61% in that region at same time.
* DB CPU 45% [normal], no saturation.
* No deploys for 6 hours to API or infra.
* One endpoint accounts for 70% of the latency increase.

**WORKING HYPOTHESES [Unconfirmed - Do not act on alone]:**
* Cache tier degradation / node loss in that region causing miss storm -> origin overload for that endpoint.
* Data / traffic shape change for that endpoint causing key cardinality blow-up / cache stampede.
* Regional capacity issue - client pool saturation, instance brownout, network partition to cache.

**NOT doing right now:** Cache flush, DB failover/restart, new deploy/rollback. All are high-risk and destructive to evidence.

---
### Top 3 Checks - In Priority Order

**1. Cache Tier Health in Affected Region - Highest signal**
Owner: Cache / SRE. Timebox: 5 min.
* Are all cache nodes members? Any node down/evicted at 14:05?
* Memory pressure, evictions/sec, OOMs, CPU, client connections, network errors.
* Client-side errors from API to cache, not just server-side.

**2. Hot Endpoint Traffic Analysis**
Owner: Backend. Timebox: 5 min.
* Traces for `/v1/<hot-endpoint>`: QPS change at 14:05? Tenant / customer spike?
* Cache key pattern: Hit rate per key, new keys per second, payload size. Is there a single hot key or key explosion?
* Origin latency without cache - is the DB / downstream for this endpoint spiking?

**3. Regional Serving Health**
Owner: SRE
* Pod/instance count, crash loops, autoscaler events, CPU throttling/GC for API fleet in that region at 14:05.
* Service mesh / load balancer latency between API and cache tier.
* Breakdown of that 0.8% errors: 504 vs 500 vs 429?

### Safe, Reversible Mitigation Proposed

Goal: Protect customers while preserving evidence.

**Proposal: Traffic shed + regional shift:**
1.  Put a **temporal rate limit / load shed** on `GET /v1/<hot-endpoint>` in affected region only - e.g., 50% of normal limit, return 429 with Retry-After. Protects other endpoints.
2.  **Shift 50% of API traffic out of affected region** to healthy regions via Global LB weight change. 1-click reversible.
3.  Enable request coalescing + stale-while-revalidate for that endpoint if feature-flag exists.

**Why safe:** No data loss, no restarts, instantly reversible via config/LB. Directly addresses symptom - cache miss storm - without touching DB/cache.

**Rollback:** Revert LB weight + remove rate limit.

Next update in 20 minutes or upon completion of check #1.

Please attach cache and endpoint dashboards in thread.

{
  "id": "resp_6ac7b01785cd56aad50246a5",
  "object": "response",
  "created_at": 1791471639,
  "model": "muse-spark-1.1",
  "output": [
    {
      "type": "reasoning",
      "id": "rs_6ac7b01785cd56aad50246a5:rs_322771d5-5cdd-4e2d-a99d-9f26fb95b8a9",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "message",
      "id": "msg_f39c4926-6745-4033-90d7-961b45acd92b",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "### INCIDENT UPDATE #1 - 14:20 UTC - API Latency Elevation [Single Region]\n\n**Status:** Investigating - SEV2\n**Impact:** Customers served from `us-east-1 [example]` experiencing API p95 of 2.4s vs normal 180ms since 14:05 UTC. p50 stable. Error rate still low at 0.8% but we are seeing timeouts. Other regions healthy.\n**Scope:** ~70% of added latency is isolated to a single endpoint: `GET /v1/<hot-endpoint>`\n\n**FACTS:**\n* 14:05 UTC start, sustained. Single region only.\n* p95: 180ms -> 2.4s. Error rate: 0.8%\n* Cache hit rate: 92% -> 61% in that region at same time.\n* DB CPU 45% [normal], no saturation.\n* No deploys for 6 hours to API or infra.\n* One endpoint accounts for 70% of the latency increase.\n\n**WORKING HYPOTHESES [Unconfirmed - Do not act on alone]:**\n* Cache tier degradation / node loss in that region causing miss storm -> origin overload for that endpoint.\n* Data / traffic shape change for that endpoint causing key cardinality blow-up / cache stampede.\n* Regional capacity issue - client pool saturation, instance brownout, network partition to cache.\n\n**NOT doing right now:** Cache flush, DB failover/restart, new deploy/rollback. All are high-risk and destructive to evidence.\n\n---\n### Top 3 Checks - In Priority Order\n\n**1. Cache Tier Health in Affected Region - Highest signal**\nOwner: Cache / SRE. Timebox: 5 min.\n* Are all cache nodes members? Any node down/evicted at 14:05?\n* Memory pressure, evictions/sec, OOMs, CPU, client connections, network errors.\n* Client-side errors from API to cache, not just server-side.\n\n**2. Hot Endpoint Traffic Analysis**\nOwner: Backend. Timebox: 5 min.\n* Traces for `/v1/<hot-endpoint>`: QPS change at 14:05? Tenant / customer spike?\n* Cache key pattern: Hit rate per key, new keys per second, payload size. Is there a single hot key or key explosion?\n* Origin latency without cache - is the DB / downstream for this endpoint spiking?\n\n**3. Regional Serving Health**\nOwner: SRE\n* Pod/instance count, crash loops, autoscaler events, CPU throttling/GC for API fleet in that region at 14:05.\n* Service mesh / load balancer latency between API and cache tier.\n* Breakdown of that 0.8% errors: 504 vs 500 vs 429?\n\n### Safe, Reversible Mitigation Proposed\n\nGoal: Protect customers while preserving evidence.\n\n**Proposal: Traffic shed + regional shift:**\n1.  Put a **temporal rate limit / load shed** on `GET /v1/<hot-endpoint>` in affected region only - e.g., 50% of normal limit, return 429 with Retry-After. Protects other endpoints.\n2.  **Shift 50% of API traffic out of affected region** to healthy regions via Global LB weight change. 1-click reversible.\n3.  Enable request coalescing + stale-while-revalidate for that endpoint if feature-flag exists.\n\n**Why safe:** No data loss, no restarts, instantly reversible via config/LB. Directly addresses symptom - cache miss storm - without touching DB/cache.\n\n**Rollback:** Revert LB weight + remove rate limit.\n\nNext update in 20 minutes or upon completion of check #1.\n\nPlease attach cache and endpoint dashboards in thread.",
          "annotations": [],
          "logprobs": []
        }
      ],
      "status": "completed"
    }
  ],
  "status": "completed",
  "usage": {
    "input_tokens": 146,
    "output_tokens": 2500,
    "total_tokens": 2646,
    "input_tokens_details": {
      "cached_tokens": 0
    },
    "output_tokens_details": {
      "reasoning_tokens": 1714
    }
  },
  "top_logprobs": 0,
  "temperature": 1,
  "top_p": 1,
  "service_tier": "auto",
  "reasoning": {
    "effort": "high"
  },
  "background": false,
  "tools": [],
  "tool_choice": "auto",
  "truncation": "disabled",
  "completed_at": 1791471639,
  "error": null,
  "incomplete_details": null,
  "instructions": null,
  "parallel_tool_calls": true,
  "store": true,
  "max_output_tokens": 4096
}

Web Search Grounding — Use Meta's Responses API web search grounding to answer a current-events question with citations.

const response = await env.AI.run(
  'meta/muse-spark-1.1',
  {
    input: 'What is the latest major technology news today? Cite the sources.',
    tools: [{ type: 'web_search' }],
  },
)
console.log(response)

curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/run \
  --header "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "meta/muse-spark-1.1",
  "input": {
    "input": "What is the latest major technology news today? Cite the sources.",
    "tools": [{"type": "web_search"}]
  }
}'

Got it — you want today’s major tech headlines with sources. I’ll pull the latest from October 8 and verify the top stories.

{
  "id": "resp_6ac7b0173b4b3903c1914984",
  "object": "response",
  "created_at": 1791471639,
  "model": "muse-spark-1.1",
  "output": [
    {
      "type": "reasoning",
      "id": "rs_6ac7b0173b4b3903c1914984:rs_4cb6db0b-4ac9-4786-8fc1-d448358e2317",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "message",
      "id": "rs_4bc20bd7-7650-465e-9a9e-ecb58e38b16d",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Got it — you want today’s major tech headlines with sources. I’ll pull the latest from October 8 and verify the top stories.",
          "annotations": [],
          "logprobs": []
        }
      ],
      "phase": "commentary",
      "status": "completed"
    },
    {
      "type": "web_search_call",
      "id": "ws_call_9345dd9b-7234-4e0c-a576-b98a32283cc5",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "latest major technology news today October 8 2026"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_0fbf1dcd-212c-4e2b-8d1e-da68d686a4ee",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "top technology headlines today"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_43240af6-97be-4ef3-9fb4-8ad53b4a40f2",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "breaking tech news October 8 2026"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_bf3ee591-c908-4e45-b400-aea7d24fda29",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "major tech news today Wall Street Journal NYTimes Bloomberg"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_22a2bb67-1e1d-49f0-b8b4-4f809ae09b1b",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "Google OpenAI Apple news October 8"
      }
    },
    {
      "type": "reasoning",
      "id": "rs_6ac7b0173b4b3903c1914984:rs_7daa6a64-a490-4064-bc3d-4a0f5ab86650",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "web_search_call",
      "id": "ws_call_804551ed-7be5-4670-8f58-1913d4656e50",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "The Verge tech news October 8 2026"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_e354cf9c-bf87-416d-910e-636fe80b44f8",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "TechCrunch headlines October 8 2026"
      }
    },
    {
      "type": "reasoning",
      "id": "rs_6ac7b0173b4b3903c1914984:rs_b8ffbebf-ce96-47ac-9823-7455a4f5a408",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "message",
      "id": "rs_c3c4da30-d286-432f-a60c-bfce20ca5d7e",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "I’m seeing early signals like Google’s SynthID rollout today — let me dig deeper into the major outlets to finalize the top stories.",
          "annotations": [],
          "logprobs": []
        }
      ],
      "phase": "commentary",
      "status": "completed"
    },
    {
      "type": "web_search_call",
      "id": "ws_call_34cb8e12-7f7a-4f0d-a1bf-bbfced64f594",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "Reuters technology news today October 8"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_1d8eb5fa-cadf-459b-b950-2a665bb1c76e",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "Bloomberg technology news October 8 2026 AI"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_732b9e15-5131-4df6-86e2-bb01fc03aa47",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "Microsoft Google Amazon tech announcement October 8 2026"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_9f5bc14b-3970-42e4-aaa3-e3d72247602d",
      "status": "completed",
      "action": {
        "type": "open_page",
        "url": "https://nokiamob.net/2026/10/08/google-opens-synthid-detector-to-everyone-heres-what-it-actually-checks/"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_b200e4a3-1c14-43d5-9161-1aefab2f0434",
      "status": "completed",
      "action": {
        "type": "open_page",
        "url": "https://winbuzzer.com/2026/10/08/google-opens-synthid-ai-watermark-checker-to-everyone-xcxwbn/"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_bdc61c33-248d-4023-933f-41744138fa9b",
      "status": "completed",
      "action": {
        "type": "open_page",
        "url": "https://kelofm.com/2026/10/08/tencent-mulls-5-billion-bond-sale-to-push-ai-ambitions-bloomberg-news-reports/"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_5f4b1af6-14be-43db-b742-50ad5dbb3ff0",
      "status": "completed",
      "action": {
        "type": "open_page",
        "url": "https://www.tradingview.com/news/coinmarketcal:16307c967094b:0-arcium-confidential-ai-removes-the-cpu-enclave-08-oct-2026/"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_8d2aea46-4bc9-4180-8ad4-d627006dedca",
      "status": "completed",
      "action": {
        "type": "open_page",
        "url": "https://mspoweruser.com/hr/heres-your-reminder-that-today-october-8-is-the-last-day-microsoft-supports-windows-11-21h2/"
      }
    },
    {
      "type": "reasoning",
      "id": "rs_6ac7b0173b4b3903c1914984:rs_e958b0b2-1277-4682-97b9-ae809a897a76",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "web_search_call",
      "id": "ws_call_cc1d22f5-d4e9-406c-b0dc-aec40cf9447e",
      "status": "completed",
      "action": {
        "type": "search",
        "query": "Apple announces October 13 event home hub HomePod mini Apple TV"
      }
    },
    {
      "type": "web_search_call",
      "id": "ws_call_c5a6ebbc-a369-4117-bf73-2955b9375902",
      "status": "completed",
      "action": {
        "type": "open_page",
        "url": "https://www.macrumors.com/2026/10/04/apple-event-reportedly-planned-for-october-13/"
      }
    },
    {
      "type": "reasoning",
      "id": "rs_6ac7b0173b4b3903c1914984:rs_11eae94f-32d9-4ea4-8712-3606c0b3b9d9",
      "summary": [],
      "status": "completed"
    },
    {
      "type": "message",
      "id": "msg_397cf2c5-50f7-4b76-88ae-bf3ea2b79bf7",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Based on today's coverage **October 8, 2026**, the major technology headlines are:\n\n### 1. Google opens AI watermark detector to everyone\nGoogle made its SynthID Detector available globally in English today.\n\n*   Google has made SynthID Detector available worldwide in English. The tool lets people check images, videos and audio files for hidden AI watermarks\n*   Coverage: It recognizes supported Google, OpenAI, NVIDIA and Kakao watermarks, with Apple detection coming later\n*   The standalone detector is available globally in English as of October 7\n\nThe tool is seen as a major step for AI transparency, with checks for invisible watermarks identifying supported AI-generated or edited media.\n\n### 2. Tencent mulling $5B bond sale to fund AI push\nChinese tech giant Tencent is raising debt to fund AI infrastructure:\n\n> Tencent is considering raising up to $5 billion via an offshore bond sale, Bloomberg News reported on Thursday\n\n### 3. Apple reportedly planning smart-home event Oct. 13\nPer *Bloomberg*'s Mark Gurman:\n\n*   Apple plans to unveil an all-new smart home hub, a new HomePod mini, and a new Apple TV 4K on Tuesday, October 13\n*   Apple is planning to hold an event that day \n*   Earlier Gurman reporting also not
... response truncated ...

Parameters

▶input

one ofrequired

instructions

string

temperature

numberminimum: 0maximum: 2

max_output_tokens

numberexclusiveMinimum: 0

top_p

numberminimum: 0maximum: 1

stream

boolean

▶tools[]

array

tool_choice

▶text{}

object

▶reasoning{}

object

string

object

stringconst: response

created_at

number

model

string

▶output[]

array

output_text

string

status

stringenum: in_progress, completed, failed, incomplete

▶usage{}

object

API Schemas (Raw)

Input

Output