WonderFull
A large language model released by Meta for agent scenarios, used in MuseCoding, with notable performance on intelligence indices and coding agent benchmarks.
Core Facts
In the Intelligence Index given by the independent evaluation organization Artificial Analysis, Muse Wonder scores only 52, while Claude Opus 5.5 reaches 58.
In the DeepSWE 1.1 benchmark, Muse Wonder Full achieved 79.0%, exceeding Claude Opus 5.5's 74.2% and GPT-6 Astra's 74.1%.
Timeline (last 90 days)
AI researcher Mehu Mohan said he was also dissatisfied with WonderFull, believing it is far inferior to Opus 5.5 and even worse than GPT-6.1 Sol.
On Terminal Bench 4.0, WonderFull trails the frontier competitors by only 2 points.
WonderFull ranks first on Meta's internal "Personal Superintelligence Index," leading Astra and Opus by about 4 percentage points.
WonderFull reaches state-of-the-art levels on the Harvey Lab legal work benchmark, Val's financial agent, and Automation Bench.
Developer Theo placed WonderFull in his S-tier models, tied with Opus 5.5.
Conductor's Charlie Holtz called WonderFull "found real security vulnerabilities in the codebase that other agents missed, in one go."
The Muse Wonder series includes two model variants: Wonder (a streamlined version optimized for speed) and WonderFull (the full version provided via API).
In the DeepSWE 1.1 benchmark, Muse Wonder Full achieved 79.0%, exceeding Claude Opus 5.5's 74.2% and GPT-6 Astra's 74.1%.
Meta released the MuseWonder and WonderFull model series; Wonder is already available on the Meta model API, and WonderFull is used for MuseCoding.
According to Artificial Analysis evaluation, WonderFull scores 56 on the intelligence index, a significant improvement over the previous 48; its coding Agent score is 68.
2 more timeline events
All Facts (12)
In the Intelligence Index given by the independent evaluation organization Artificial Analysis, Muse Wonder scores only 52, while Claude Opus 5.5 reaches 58.
70%Partially VerifiedIn the DeepSWE 1.1 benchmark, Muse Wonder Full achieved 79.0%, exceeding Claude Opus 5.5's 74.2% and GPT-6 Astra's 74.1%.
65%UnverifiedAI researcher Mehu Mohan said he was also dissatisfied with WonderFull, believing it is far inferior to Opus 5.5 and even worse than GPT-6.1 Sol.
50%UnverifiedWonderFull reaches state-of-the-art levels on the Harvey Lab legal work benchmark, Val's financial agent, and Automation Bench.
50%UnverifiedDeveloper Theo placed WonderFull in his S-tier models, tied with Opus 5.5.
50%UnverifiedConductor's Charlie Holtz called WonderFull "found real security vulnerabilities in the codebase that other agents missed, in one go."
50%UnverifiedThe Muse Wonder series includes two model variants: Wonder (a streamlined version optimized for speed) and WonderFull (the full version provided via API).
50%UnverifiedOn Terminal Bench 4.0, WonderFull trails the frontier competitors by only 2 points.
50%UnverifiedWonderFull ranks first on Meta's internal "Personal Superintelligence Index," leading Astra and Opus by about 4 percentage points.
50%UnverifiedWonderFull's coding Agent score of 68 is on par with ClawedCode (equipped with Sonnet 5.5).
50%UnverifiedAccording to Artificial Analysis evaluation, WonderFull scores 56 on the intelligence index, a significant improvement over the previous 48; its coding Agent score is 68.
50%UnverifiedMeta released the MuseWonder and WonderFull model series; Wonder is already available on the Meta model API, and WonderFull is used for MuseCoding.
50%