GPT-5.6 Luna vs Muse Wonder
Muse Wonder leads GPT-5.6 Luna by 9.5 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. GPT-5.6 Luna wins 0 and Muse Wonder wins 12 of 13 shared benchmarks. GPT-5.6 Luna is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | GPT-5.6 Luna | Muse Wonder |
|---|---|---|
| Agentic coding | 59 | 68 |
| Long-horizon autonomy | 61 | 70 |
| Knowledge-work agents | 39 | 50 |
| Computer use | 40 | 50 |
| Security & context learning | 51 | 57 |
Shared benchmarks
| Benchmark | GPT-5.6 Luna | Muse Wonder |
|---|---|---|
| APEX-Agentsindex | 43.0% | 66.1% |
| Vending-Bench 2index | $4,095 | $9,432 |
| CursorBenchindex | 35.9% | 50.4% |
| CritPt | 20.6% | 28.3% |
| DTBench | 89.1% | 95.2% |
| FrontierMath Tier 4 | 61.0% | 70.7% |
| FrontierMath Tiers 1-3 | 82.1% | 82.1% |
| LMCA | 48.5% | 56.8% |
| MirrorCode | 21.1% | 45.6% |
| OTIS Mock AIME | 98.3% | 99.4% |
| ProofBench | 60.0% | 84.0% |
| SciCode | 53.6% | 60.3% |
| WeirdML v3 | 5.2% | 15.1% |
Specs and price
| GPT-5.6 Luna | Muse Wonder | |
|---|---|---|
| Agent Index | 51.4 (#34) | 60.9 (#15) |
| Agent $/M | $0.10 | $1.69 |
| Input / output $/M | $0.20 / $1.20 | $4.00 / $20.00 |
| Context | 1.1M | 1M |
| Released | Jul 9, 2026 | Oct 12, 2026 |
| Weights | Closed | Closed |
| API hosts | 5 | 2 |
Don’t pick one. Route.
Shogo can run GPT-5.6 Luna and Muse Wonder side by side and send each step of an agent to the one that’s cheaper for it.
Try both in Shogo