Claude Fable 5.2 vs GPT-5.6 Luna
Claude Fable 5.2 leads GPT-5.6 Luna by 25.0 points on the Shogo Agent Index. Claude Fable 5.2 wins 18 and GPT-5.6 Luna wins 0 of 18 shared benchmarks. GPT-5.6 Luna is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | Claude Fable 5.2 | GPT-5.6 Luna |
|---|---|---|
| Agentic coding | 86 | 59 |
| Long-horizon autonomy | 78 | 61 |
| Knowledge-work agents | 67 | 39 |
| Computer use | 77 | 40 |
| Security & context learning | 65 | 51 |
Shared benchmarks
| Benchmark | Claude Fable 5.2 | GPT-5.6 Luna |
|---|---|---|
| APEX-Agentsindex | 80.1% | 43.0% |
| Vending-Bench 2index | $9,628 | $4,095 |
| CursorBenchindex | 60.0% | 35.9% |
| ARC-AGI | 98.5% | 88.0% |
| ARC-AGI-2 | 93.8% | 59.5% |
| Chess Puzzles | 79.0% | 43.0% |
| CritPt | 35.7% | 20.6% |
| DTBench | 99.2% | 89.1% |
| FrontierMath Tier 4 | 100.0% | 61.0% |
| FrontierMath Tiers 1-3 | 95.1% | 82.1% |
| GPQA Diamond | 96.0% | 91.6% |
| LMCA | 70.6% | 48.5% |
| OTIS Mock AIME | 99.7% | 98.3% |
| ProofBench | 100.0% | 60.0% |
| SciCode | 67.7% | 53.6% |
| SimpleBench | 94.9% | 46.8% |
| SimpleQA Verified | 94.0% | 41.0% |
| WeirdML v3 | 43.1% | 5.2% |
Specs and price
| Claude Fable 5.2 | GPT-5.6 Luna | |
|---|---|---|
| Agent Index | 76.4 (#1) | 51.4 (#34) |
| Agent $/M | $3.21 | $0.10 |
| Input / output $/M | $8.00 / $40.00 | $0.20 / $1.20 |
| Context | 1M | 1.1M |
| Released | Nov 9, 2026 | Jul 9, 2026 |
| Weights | Closed | Closed |
| API hosts | 4 | 5 |
Don’t pick one. Route.
Shogo can run Claude Fable 5.2 and GPT-5.6 Luna side by side and send each step of an agent to the one that’s cheaper for it.
Try both in Shogo