Claude Sonnet 5.5 vs GPT-5.6 Terra
Claude Sonnet 5.5 leads GPT-5.6 Terra by 10.2 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. Claude Sonnet 5.5 wins 18 and GPT-5.6 Terra wins 1 of 20 shared benchmarks. Claude Sonnet 5.5 is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | Claude Sonnet 5.5 | GPT-5.6 Terra |
|---|---|---|
| Agentic coding | 79 | 66 |
| Long-horizon autonomy | 75 | 69 |
| Knowledge-work agents | 57 | 47 |
| Computer use | 61 | 48 |
| Security & context learning | 61 | 56 |
Shared benchmarks
| Benchmark | Claude Sonnet 5.5 | GPT-5.6 Terra |
|---|---|---|
| APEX-Agentsindex | 75.5% | 58.2% |
| Vending-Bench 2index | $11,165 | $7,343 |
| CursorBenchindex | 55.5% | 41.3% |
| ARC-AGI | 96.5% | 96.5% |
| ARC-AGI-2 | 83.9% | 83.9% |
| Chess Puzzles | 59.0% | 47.0% |
| CritPt | 31.4% | 30.0% |
| DTBench | 97.3% | 93.3% |
| FrontierCode | 52.1% | 41.3% |
| FrontierMath Tier 4 | 80.5% | 70.7% |
| FrontierMath Tiers 1-3 | 88.8% | 86.0% |
| GPQA Diamond | 95.6% | 93.3% |
| LMCA | 68.3% | 54.9% |
| MirrorCode | 97.8% | 41.1% |
| OTIS Mock AIME | 100.0% | 99.7% |
| ProofBench | 100.0% | 74.0% |
| SciCode | 61.0% | 55.0% |
| SimpleBench | 75.9% | 48.9% |
| SimpleQA Verified | 46.5% | 43.2% |
| WebDev Arena | 1786 | 1521 |
Specs and price
| Claude Sonnet 5.5 | GPT-5.6 Terra | |
|---|---|---|
| Agent Index | 69 (#9) | 58.8 (#16) |
| Agent $/M | $0.84 | $1.02 |
| Input / output $/M | $2.00 / $10.00 | $2.00 / $12.00 |
| Context | 1M | 1.1M |
| Released | Sep 28, 2026 | Jul 9, 2026 |
| Weights | Closed | Closed |
| API hosts | 5 | 4 |
Don’t pick one. Route.
Shogo can run Claude Sonnet 5.5 and GPT-5.6 Terra side by side and send each step of an agent to the one that’s cheaper for it.
Try both in Shogo