{
  "api_version": "1",
  "updated": "2026-11-10",
  "latest_event": "2026-11-09",
  "rights": "© Frontier Log, all rights reserved.",
  "event": {
    "slug": "2026-11-09-muse-ember",
    "date": "2026-11-09",
    "provider": "meta",
    "importance": "minor",
    "title": "Muse Ember",
    "headline": "Meta launches Muse Ember",
    "detail": "Meta's smallest and most efficient model, for the work agents repeat all day, on the Meta Model API at $0.10 and $0.50 per million input and output tokens.",
    "tags": [
      "model-releases",
      "agents",
      "small-models"
    ],
    "related": [
      {
        "slug": "2026-10-12-muse-wonder",
        "date": "2026-10-12",
        "provider": "meta",
        "title": "Muse Wonder and Muse Wonder Full",
        "headline": "Meta launches Muse Wonder and Muse Wonder Full",
        "url": "https://frontierlog.ai/event/2026-10-12-muse-wonder"
      }
    ],
    "source": "https://research.meta.ai/blog/muse-ember",
    "models": [
      {
        "slug": "muse-ember",
        "name": "Muse Ember",
        "lab": "meta",
        "date": "2026-11-09",
        "source": "https://research.meta.ai/blog/muse-ember",
        "source_title": "Introducing Our Smallest, Most Efficient Model: Muse Ember",
        "flagship": false,
        "event": "https://frontierlog.ai/event/2026-11-09-muse-ember",
        "releases": [],
        "variant_of": null,
        "made_by": "Meta",
        "url": "https://frontierlog.ai/model/muse-ember",
        "results": [
          {
            "id": "muse-ember:automationbench-1-0-6:aa:1",
            "model": "muse-ember",
            "model_name": "Muse Ember",
            "model_date": "2026-11-09",
            "event": "https://frontierlog.ai/event/2026-11-09-muse-ember",
            "benchmark": "automationbench-1-0-6",
            "benchmark_name": "AutomationBench 1.0.6",
            "setting": "aa",
            "setting_name": "AutomationBench-AA",
            "score": 37.8,
            "unit": "%",
            "printed": "37.8",
            "reported_by": "meta",
            "run_by": "Artificial Analysis",
            "label": "Muse Ember (max)",
            "source": "https://research.meta.ai/blog/muse-ember",
            "source_title": "Introducing Our Smallest, Most Efficient Model: Muse Ember",
            "source_date": "2026-11-09",
            "where": "'Muse Ember evaluation results' table (HTML), row 'Agents / AutomationBench-AA', column 'Muse Ember (max)'",
            "own": true,
            "primary": true,
            "parallel": false,
            "notes": "max reasoning effort; pass@1",
            "caveat": "",
            "human_verified": false,
            "human_verified_on": null,
            "check": "double_entry",
            "corroborated": false,
            "corroborated_on": null,
            "corroborated_by": []
          },
          {
            "id": "muse-ember:terminal-bench-4-0:default:1",
            "model": "muse-ember",
            "model_name": "Muse Ember",
            "model_date": "2026-11-09",
            "event": "https://frontierlog.ai/event/2026-11-09-muse-ember",
            "benchmark": "terminal-bench-4-0",
            "benchmark_name": "Terminal-Bench 4.0",
            "setting": "default",
            "setting_name": "",
            "score": 40.9,
            "unit": "%",
            "printed": "40.9",
            "reported_by": "meta",
            "run_by": "Artificial Analysis",
            "label": "Muse Ember (max)",
            "source": "https://research.meta.ai/blog/muse-ember",
            "source_title": "Introducing Our Smallest, Most Efficient Model: Muse Ember",
            "source_date": "2026-11-09",
            "where": "'Muse Ember evaluation results' table (HTML), row 'Coding / Terminal-Bench 4.0', column 'Muse Ember (max)'",
            "own": true,
            "primary": true,
            "parallel": false,
            "notes": "mini-swe-agent harness; max reasoning effort",
            "caveat": "",
            "human_verified": false,
            "human_verified_on": null,
            "check": "double_entry",
            "corroborated": false,
            "corroborated_on": null,
            "corroborated_by": []
          },
          {
            "id": "muse-ember:osworld-2-1:partial:1",
            "model": "muse-ember",
            "model_name": "Muse Ember",
            "model_date": "2026-11-09",
            "event": "https://frontierlog.ai/event/2026-11-09-muse-ember",
            "benchmark": "osworld-2-1",
            "benchmark_name": "OSWorld 2.1",
            "setting": "partial",
            "setting_name": "Full set, partial score",
            "score": 69.5,
            "unit": "%",
            "printed": "69.5",
            "reported_by": "meta",
            "run_by": null,
            "label": "Muse Ember (max)",
            "source": "https://research.meta.ai/blog/muse-ember",
            "source_title": "Introducing Our Smallest, Most Efficient Model: Muse Ember",
            "source_date": "2026-11-09",
            "where": "'Muse Ember evaluation results' table (HTML), row 'Multimodal / OSWorld 2.1', column 'Muse Ember (max)', the value marked '(partial)'",
            "own": true,
            "primary": true,
            "parallel": false,
            "notes": "max reasoning effort; mean per-task partial credit; 108 tasks; Meta's own framework",
            "caveat": "",
            "human_verified": false,
            "human_verified_on": null,
            "check": "double_entry",
            "corroborated": false,
            "corroborated_on": null,
            "corroborated_by": []
          },
          {
            "id": "muse-ember:osworld-2-1:strict:1",
            "model": "muse-ember",
            "model_name": "Muse Ember",
            "model_date": "2026-11-09",
            "event": "https://frontierlog.ai/event/2026-11-09-muse-ember",
            "benchmark": "osworld-2-1",
            "benchmark_name": "OSWorld 2.1",
            "setting": "strict",
            "setting_name": "Full set, strict pass rate",
            "score": 33.8,
            "unit": "%",
            "printed": "33.8",
            "reported_by": "meta",
            "run_by": null,
            "label": "Muse Ember (max)",
            "source": "https://research.meta.ai/blog/muse-ember",
            "source_title": "Introducing Our Smallest, Most Efficient Model: Muse Ember",
            "source_date": "2026-11-09",
            "where": "'Muse Ember evaluation results' table (HTML), row 'Multimodal / OSWorld 2.1', column 'Muse Ember (max)', the value marked '(binary)'",
            "own": true,
            "primary": true,
            "parallel": false,
            "notes": "max reasoning effort; share of tasks with full credit; 108 tasks; Meta's own framework",
            "caveat": "",
            "human_verified": false,
            "human_verified_on": null,
            "check": "double_entry",
            "corroborated": false,
            "corroborated_on": null,
            "corroborated_by": []
          }
        ],
        "price": [
          {
            "from": "2026-11-09",
            "listed": false,
            "currency": "USD",
            "input_per_1m": 0.1,
            "cached_input_per_1m": 0.004,
            "cached_input_derived": false,
            "output_per_1m": 0.5,
            "cache": {
              "kind": "price"
            },
            "tiers": [
              {
                "over": 200000,
                "measure": "prompt",
                "input_per_1m": 0.3,
                "cached_input_per_1m": 0.012,
                "cached_input_derived": false,
                "output_per_1m": 1.5,
                "printed": {
                  "input": "$0.30",
                  "cached_input": "$0.012",
                  "output": "$1.50"
                }
              }
            ],
            "printed": {
              "input": "$0.10",
              "cached_input": "$0.004",
              "output": "$0.50"
            },
            "printed_unit": "1M tokens",
            "notes": "",
            "sources": [
              "https://research.meta.ai/blog/muse-ember",
              "https://dev.meta.ai/models/muse-ember/"
            ],
            "check": "corroborated"
          }
        ],
        "context": {
          "from": "2026-11-09",
          "listed": false,
          "window_tokens": 1000000,
          "input_tokens": null,
          "output_tokens": 512000,
          "tiers": [],
          "printed": {
            "window": "1M",
            "input": null,
            "output": "512K"
          },
          "notes": "",
          "sources": [
            "https://research.meta.ai/blog/muse-ember",
            "https://dev.meta.ai/models/muse-ember/"
          ],
          "check": "corroborated"
        },
        "prices": [
          {
            "from": "2026-11-09",
            "listed": false,
            "currency": "USD",
            "input_per_1m": 0.1,
            "cached_input_per_1m": 0.004,
            "cached_input_derived": false,
            "output_per_1m": 0.5,
            "cache": {
              "kind": "price"
            },
            "tiers": [
              {
                "over": 200000,
                "measure": "prompt",
                "input_per_1m": 0.3,
                "cached_input_per_1m": 0.012,
                "cached_input_derived": false,
                "output_per_1m": 1.5,
                "printed": {
                  "input": "$0.30",
                  "cached_input": "$0.012",
                  "output": "$1.50"
                }
              }
            ],
            "printed": {
              "input": "$0.10",
              "cached_input": "$0.004",
              "output": "$0.50"
            },
            "printed_unit": "1M tokens",
            "notes": "",
            "sources": [
              "https://research.meta.ai/blog/muse-ember",
              "https://dev.meta.ai/models/muse-ember/"
            ],
            "check": "corroborated"
          }
        ],
        "context_history": [
          {
            "from": "2026-11-09",
            "listed": false,
            "window_tokens": 1000000,
            "input_tokens": null,
            "output_tokens": 512000,
            "tiers": [],
            "printed": {
              "window": "1M",
              "input": null,
              "output": "512K"
            },
            "notes": "",
            "sources": [
              "https://research.meta.ai/blog/muse-ember",
              "https://dev.meta.ai/models/muse-ember/"
            ],
            "check": "corroborated"
          }
        ],
        "stories": [],
        "compare": []
      }
    ],
    "source_title": "Introducing Our Smallest, Most Efficient Model: Muse Ember",
    "source_archived": null,
    "source_http_status": null,
    "source_quote": "Muse Ember is available today on the Meta Model API, at $0.10 per million input tokens and $0.50 per million output tokens.",
    "followups": [],
    "successor": null,
    "predecessors": [],
    "revealed_as": null,
    "stealth_previews": [],
    "reveal_source": null,
    "url": "https://frontierlog.ai/event/2026-11-09-muse-ember",
    "image": "https://frontierlog.ai/og/2026-11-09-muse-ember.png",
    "stories": []
  }
}