GPT-5.6 Luna vs GPT-6 Sol
GPT-6 Sol leads GPT-5.6 Luna by 12.2 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. GPT-5.6 Luna wins 0 and GPT-6 Sol wins 20 of 20 shared benchmarks. GPT-5.6 Luna is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | GPT-5.6 Luna | GPT-6 Sol |
|---|---|---|
| Agentic coding | 59 | 72 |
| Long-horizon autonomy | 61 | 76 |
| Knowledge-work agents | 39 | 48 |
| Computer use | 40 | 54 |
| Security & context learning | 51 | 59 |
Shared benchmarks
| Benchmark | GPT-5.6 Luna | GPT-6 Sol |
|---|---|---|
| APEX-Agentsindex | 43.0% | 54.3% |
| Vending-Bench 2index | $4,095 | $14,428 |
| ARC-AGI | 88.0% | 95.5% |
| ARC-AGI-2 | 59.5% | 89.6% |
| Chess Puzzles | 43.0% | 56.0% |
| CritPt | 20.6% | 30.9% |
| DTBench | 89.1% | 97.3% |
| FrontierCode | 39.8% | 49.3% |
| FrontierMath Tier 4 | 61.0% | 90.0% |
| FrontierMath Tiers 1-3 | 82.1% | 89.8% |
| GPQA Diamond | 91.6% | 94.3% |
| LMCA | 48.5% | 59.1% |
| MirrorCode | 21.1% | 48.9% |
| OTIS Mock AIME | 98.3% | 100.0% |
| ProofBench | 60.0% | 83.0% |
| SciCode | 53.6% | 57.6% |
| SimpleBench | 46.8% | 73.1% |
| SimpleQA Verified | 41.0% | 60.7% |
| WebDev Arena | 1519 | 1689 |
| WeirdML v3 | 5.2% | 19.7% |
Specs and price
| GPT-5.6 Luna | GPT-6 Sol | |
|---|---|---|
| Agent Index | 51.4 (#34) | 63.6 (#13) |
| Agent $/M | $0.10 | $0.92 |
| Input / output $/M | $0.20 / $1.20 | $2.00 / $10.00 |
| Context | 1.1M | 1.1M |
| Released | Jul 9, 2026 | Sep 22, 2026 |
| Weights | Closed | Closed |
| API hosts | 5 | 5 |
Don’t pick one. Route.
Shogo can run GPT-5.6 Luna and GPT-6 Sol side by side and send each step of an agent to the one that’s cheaper for it.
Try both in Shogo