Observed Signal · Oct 13, 2026 · Product Launch · Source: Trending Topics (DACH/CEE Innovation & Tech) · Impact: 3/5 · Sentiment: Positive
Meta's Muse Wonder matches Claude Fable for less
Independent benchmarkers at Artificial Analysis have analyzed Meta's new AI models, Muse Wonder and Muse Wonder Full. The Intelligence Index shows Muse Wonder Full scoring 55.8, narrowly beating Anthropic's Claude Fable 5.1 (53.4) and placing third behind Claude Opus 5.5 and Claude Sonnet 5.5, while also surpassing GPT-6 Astra (52.7). The standard Muse Wonder scores 52.4. Notably, Muse Wonder costs about 40% less per task than Claude Fable 5.1, despite a 2.8x price increase from its predecessor Muse Spark 1.3. The models excel in AI agent tasks, coding, and scientific reasoning, with reduced hallucination rates (23% for Muse Wonder Full). However, they have higher input token consumption and gaps in knowledge tests. The stronger Muse Wonder Full is currently limited to Muse Code users.
Meta's new AI models close the performance gap with leading models at a lower cost, impacting AI adoption in advertising technology and marketing automation.
MARKET SIGNALS
Track Meta Signals & Strategic Shifts
Polaris7 connects regulatory filings, primary sources, leadership changes, and M&A activity into structured intelligence. Set up your Polaris7 workspace to monitor relevant companies in your executive briefing.
- Verified primary sources
- Automated briefings
- Instant watchlist setup
Key Takeaways & Evidence Grounding
- Meta released Muse Wonder and Muse Wonder Full, scoring 52.4 and 55.8 on the Intelligence Index, with Full surpassing Claude Fable 5.1 and GPT-6 Astra.
- Muse Wonder costs about 40% less per task than Claude Fable 5.1, though it is 2.8x more expensive than its predecessor Muse Spark 1.3.
- Muse Wonder Full scores 68 on the Coding Agents Index with Muse Code, matching Claude Code with Claude Sonnet 5.5.
- Muse Wonder's hallucination rate dropped from 33% to 23% compared to Muse Spark 1.3.
- Muse Wonder processes 7.69 million input tokens per task, significantly higher than competitors, and token prices increased from $1.25/$4.25 to $4/$20 per million tokens.
Connected Companies & Entities
“Meta's new A.I. models, Muse Wonder and Muse Wonder Full, were analyzed by Artificial Analysis.”
“The independent benchmarkers at Artificial Analysis have taken a detailed look at Meta's new models.”
“Anthropic's Claude Fable 5.1 is surpassed by Muse Wonder Full in the Intelligence Index.”
“GPT-6 Astra and GPT-6.1 Sol are mentioned in the Intelligence Index and price rankings.”
“Gemini 4 Argon is referenced in the benchmark comparisons.”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Meta Unveils Frontier Models Muse Wonder
Meta Superintelligence Labs (MSL) launched its second-generation frontier AI models, Muse Wonder and Wonder Full, featuring one-million-token context windows and Active Memory/Active Focus capabilities. Independent benchmarks (Artificial Analysis Index v4.3) rank Wonder Full second (score 56) behind Claude Opus 5.5 (58); Meta claims parity with GPT-6.1 Sol and Claude Fable 5. Pricing is $4/$20 per million input/output tokens, with Wonder available via API and Muse Code, while Wonder Full is Muse Code-exclusive. Additionally, Meta released Muse Spark Open (816B parameters, 42B active), positioned as the best non-Chinese open-weight model, optimized for Apple's Mac Studio. Meta's strategic pivot includes a $160 billion investment and monetization via transaction fees. A launch-day usage-limit bug was fixed with a quota reset. In other AI news: CoreWeave faced debt concerns (shares -15.5%), Character.AI filed for Chapter 11, Modal raised $750M Series D, and Anthropic's Q3 revenue hit $16.8B.
Meta's Muse Spark 1.3 Nears Top AI Rankings
Meta released Muse Spark 1.3, its latest frontier AI model, scoring 61 on the Artificial Analysis Intelligence Index and 62 on max preview, tying or surpassing GPT-5.6 Sol, Grok 4.6, and Claude Opus 5, with only Anthropic's Claude Fable 5.1 and Claude Opus 5 ranking higher. Gains stem from improved agentic tasks and scientific reasoning. Pricing is aggressive at $0.55 per Intelligence Index task, while token costs remain unchanged from version 1.2. The model is available in Muse Code and via API. Meta committed to open-weight releases of Muse Spark 1.2, but 1.3 remains undecided, potentially impacting the open-source ecosystem. Additionally, OpenAI's GPT-6 Astra scores 61 but carries significantly higher API costs.
Meta launches Muse Ember, cheap model for agent execution
Meta Superintelligence Lab released Muse Ember on Monday, its lightest and most cost-efficient AI model yet. Priced at 10 cents per million input tokens, Ember is cheaper than frontier competitors like Anthropic's Haiku and OpenAI's Luna, while also being cost-competitive with open-source models like DeepSeek V4.1 Flash and GLM 5.3 Flash. Designed for agentic workloads, Ember handles tool calls, coding subtasks, routing, extraction, and long document parsing. It is intended to pair with larger models like Muse Wonder in an agentic harness: Muse Wonder plans, while Muse Ember handles the volume. The model also processes text, images, and audio, enabling agents to read screenshots, charts, and scanned documents, or work from voice memos and meeting recordings. Ember is the fourth model in Meta's Muse family, following Spark, Glimmer, and Wonder.
WATCHLIST & BRIEFINGS
Receive future signals for Meta & competitors in your briefing
Stay ahead of material M&A deals, leadership changes, account wins, and product rollouts. Polaris7 connects primary filings, company newsrooms, and trade reporting directly into your executive briefing.
- Regular primary-source checks
- Structured Watchlist briefings
- Preserves current context
Market Signal Monitoring
Follow affected companies in Polaris7 with regular source checks and structured Watchlist briefings.
Monitor Meta
Follow affected companies in Polaris7 with regular source checks and structured Watchlist briefings.
