Muse Wonder Full (AI model): news timeline & CVEs

Show HN: Gemini 4 Argon plays StarCraft

StarSkirmish v0.2 scores how strong a StarCraft bot each flagship LLM can write in one hour.

StarSkirmish Bench v0.2 gives each LLM one hour in its vendor CLI to write a Protoss bot in C++ for StarCraft: Brood War using BWAPI 4.4.0 and OpenBW. Scores are average win rates over five runs against 10 competitive human-written bots and three demo bots. GPT-6 Astra leads, GPT-6.1 Sol is second at roughly a tenth of Astra's API cost, and Gemini 4 Argon, Claude Opus 5.5 and Claude Fable 5.1 cluster between 37 and 39. Even Astra wins only about 7% of games against Stardust, PurpleWave and BananaBrain.

Hacker News · securityupdated Oct 16, 15:03 UTC · 9h agofirst Oct 16, 14:18 UTC · 10h agoAI research 5 sources HN 23↑ · 9 comments2· 1 read

Muse Spark Open is Meta's answer to Qwen, GLM, and Kimi in the open-weight race

Meta released Muse Spark Open, an 816-billion-parameter open-weight model for coding, agents, and multimodal tasks.

Meta released Muse Spark Open, an open-weight model it calls the most capable US open-weight model, aimed at coding and agentic work. It has 816 billion parameters with 42 billion active, up to one million tokens of context, and text, image, video, and audio support. Meta says it leads Qwen 3.8, GLM 5.3, and Kimi K3 on agentic, knowledge, reasoning, and multimodal benchmarks but trails some rivals on coding and remains well behind paid Muse Wonder Full models. Weights are on Hugging Face under a free license that requires a "Built with Muse Spark Open" label; large API providers need a separate agreement.

The Decoderupdated Oct 16, 04:16 UTC · 20h agofirst Oct 15, 18:34 UTC · 1d agoModel release 7 sources2· 2 reads

The AI race is back on

Google’s Gemini 4 Argon and Meta’s Muse Wonder Full reportedly match leading OpenAI and Anthropic models.

The Verge argues the frontier race reopened after Google released Gemini 4 Argon and Meta released Muse Wonder Full, both claimed to match OpenAI and Anthropic flagships. Google is pricing Argon as much as 80 percent cheaper per token during an introductory period. Meta says Muse Wonder Full competes with GPT-6 Astra, Claude Opus 5.5, and Claude Fable 5.1, after Muse Spark fell short. The piece also cites DeepMind leadership changes and Meta’s spending, including on Scale AI.

The Verge · AIOct 14, 19:06 UTC · 2d agoAI industry 8 sources1· 1 read

Meta Launches Muse Wonder and Muse Wonder Full: Benchmarks, Pricing and How to Use It

A headline says Meta launched Muse Wonder and Muse Wonder Full, but the page gives no details.

A MarkTechPost headline says Meta launched Muse Wonder and Muse Wonder Full and promises benchmarks, pricing, and usage instructions. The provided page text is only a cookie-consent notice and does not include model sizes, benchmark scores, prices, or release details. Those claims cannot be verified from the available text.

MarkTechPostupdated Oct 14, 19:06 UTC · 2d agofirst Oct 13, 22:53 UTC · 3d agoModel release 8 sources1

[AINews] Meta Muse Wonder & Wonder Full: MSL’s Claude 5.5-tier Frontier Model

Meta Superintelligence Labs launched Muse Wonder and Muse Wonder Full, frontier models scoring near Claude Opus 5.5.

Meta Superintelligence Labs released Muse Wonder and Muse Wonder Full, its second-generation frontier family, about 14 months after the lab was founded. Artificial Analysis Intelligence Index v4.3 scored Wonder Full (max) at 56 and Wonder (max) at 52, versus Claude Opus 5.5 at 58 and Muse Spark 1.3 at 48. Meta-reported results include SWE-Marathon at 57.5% for Wonder Full, ahead of Opus 5.5 at 46.3% and GPT-6 Astra at 42.5%. The models use a 1 million-token context, with reported pricing of $4 per million input tokens and $20 per million output tokens; a same-day Muse Code quota bug was fixed.

Latent Spaceupdated Oct 14, 19:06 UTC · 2d agofirst Oct 13, 09:37 UTC · 3d agoModel release 8 sources2· 1 read

llm-meta-ai 0.2

Meta released Muse Wonder and Muse Wonder Full, rated near GPT-6.1 Sol, with an llm plugin update.

Meta AI released Muse Wonder and Muse Wonder Full, described as its most capable models and possibly its first Fable-class systems. Artificial Analysis rates Muse Wonder (Max) as equivalent to GPT-6.1 Sol. Simon Willison updated the llm-meta-ai plugin to 0.2 to call the models, which expose six reasoning levels from minimal to max. He compared outputs with Muse 1.3, released September 2, and Muse 1.2, released August 5.

Simon Willisonupdated Oct 14, 19:06 UTC · 2d agofirst Oct 12, 23:58 UTC · 4d agoModel release 8 sources1· 1 read

Meta's new frontier model Muse Wonder Full breaks into the AI top tier

Meta released Muse Wonder and Muse Wonder Full, scoring 52 and 56 on the Artificial Analysis Intelligence Index.

Meta released Muse Wonder and Muse Wonder Full, scoring 52 and 56 on the Artificial Analysis Intelligence Index v4.3, up from 48 for Muse Spark 1.3. Muse Wonder Full places in the top three on six of ten benchmarks, ties Claude Code with Sonnet 5.5 at 68 on the Coding Agents Index, and solved 94 of 100 ProofBench problems. Both support a one-million-token context and text, image, video, and audio input; Wonder is on Meta's API and Muse Code, while Wonder Full is limited to Muse Code. Input and output prices rose to $4 and $20 per million tokens, and Artificial Analysis prices Wonder at $4.52 per Intelligence Index task.

The Decoderupdated Oct 14, 19:06 UTC · 2d agofirst Oct 12, 18:48 UTC · 4d agoModel release 8 sources7· 3 reads

Meta releases its first frontier model, Muse Wonder

Meta released frontier models Muse Wonder and Muse Wonder Full, claiming parity with Fable 5.1 and GPT-6 Astra.

Meta's Superintelligence Lab released Muse Wonder and Muse Wonder Full, its first frontier models, trained at the Prometheus data center in Ohio. Company testing claims scores comparable to Fable 5.1 and GPT-6 Astra, with higher MemoryBench results than Fable, Opus, or Astra. Through the Meta Model API, Wonder costs $4 per million input tokens and $20 per million output tokens; Wonder Full is limited to Muse Code subscribers. Both models will power the Muse agent and emphasize long-term "active memory."

TechCrunch · AIupdated Oct 14, 19:06 UTC · 2d agofirst Oct 12, 17:56 UTC · 4d agoModel release 8 sources2

Introducing Muse Wonder and Muse Wonder Full: Frontier Models for Agents

Meta released Muse Wonder and Muse Wonder Full, frontier agent models with memory, persistence, and safety limits.

Meta introduced Muse Wonder and Muse Wonder Full, calling them its most capable models and successors to Muse Spark, with state-of-the-art results on personal and professional agent tasks in law, finance, health, and coding. The models add Active Memory, which carries relevant details across conversations, and Active Focus, which persists through difficult tasks, tries alternatives, recovers from errors, and checks before important actions. Meta says they are trained not to share private information without permission, to refuse harmful requests, and to remain usable for defensive code review. Muse Wonder is available now through the Meta Model API and Muse Code, with a broader Muse subscriber rollout planned and a fuller safety report promised.

Meta Newsroomupdated Oct 14, 19:06 UTC · 2d agofirst Oct 12, 17:08 UTC · 4d agoModel release 8 sources7· 4 reads

Meta has some Wonder Full new AI models

Meta launched Muse Wonder and Muse Wonder Full, claiming parity with leading OpenAI and Anthropic models.

The Verge reports that Meta launched Muse Wonder and Muse Wonder Full after Muse Spark and the Muse agent, claiming the new models can rival leading systems from OpenAI and Anthropic. Meta points to internal tests against GPT-6 Astra, Claude Opus 5.5, and Claude Fable 5.1, while comparison with Google’s Gemini 4 Argon remains unclear. Wonder is available through the Meta Model API, and both models are available to Muse Code subscribers starting today. Meta plans to power the Muse agent with Wonder beginning with subscribers and to build specialized models, but it gave no timeline or further specifics.

The Verge · AIupdated Oct 14, 19:06 UTC · 2d agofirst Oct 12, 17:00 UTC · 4d agoModel release 8 sources4· 2 reads

Muse Wonder Full

Timeline10 articles

Appears with