Claude Sonnet 5vsGPT-5.6 Terra
Across 7 shared benchmarks, GPT-5.6 Terra leads overall: Claude Sonnet 5 wins 1, GPT-5.6 Terra wins 6, with 0 ties and an average score difference of -10.59.
Claude Sonnet 5
Anthropic · 2026-06-30 · Multimodal model
GPT-5.6 Terra
OpenAI · 2026-06-26 · Reasoning model
Claude Sonnet 51 win(14%)(86%)6 winsGPT-5.6 Terra
Benchmark scores
Grouped by capability, sorted by largest gap within each. 7 shared benchmarks.
Coding and Software Engineer
GPT-5.6 Terra 2/3| Benchmark | Claude Sonnet 5 | GPT-5.6 Terra | Diff |
|---|---|---|---|
| Text Arena (Coding) | 1,54410 / 35Thinking High (No Tools) | 1,52216 / 35极高强度思考（无工具） | +22.55 |
| DeepSWE | 5416 / 27Deep Thinking (With Tools) | 69.603 / 27极高强度思考（工具） | -15.60 |
| WeirdML v2 | 68.7820 / 52Thinking High (With Tools) | 78.2710 / 52Thinking High (With Tools) | -9.49 |
Math and Reasoning
GPT-5.6 Terra 2/2| Benchmark | Claude Sonnet 5 | GPT-5.6 Terra | Diff |
|---|---|---|---|
| FrontierMath Tier 4 v2 | 29.2716 / 34最高（无工具） | 70.736 / 34最高（无工具） | -41.46 |
| FrontierMath v2 | 65.6117 / 34最高（无工具） | 85.964 / 34最高（无工具） | -20.35 |
AI Agent - Tool Usage
GPT-5.6 Terra 1/1| Benchmark | Claude Sonnet 5 | GPT-5.6 Terra | Diff |
|---|---|---|---|
| Terminal-Bench 2.1 | 80.4018 / 44极高强度思考（工具） | 87.407 / 44最高（无工具） | -7 |
General Evaluation
GPT-5.6 Terra 1/1| Benchmark | Claude Sonnet 5 | GPT-5.6 Terra | Diff |
|---|---|---|---|
| GPQA Diamond | 90.5333 / 226极高强度思考（无工具） | 93.3116 / 226最高（无工具） | -2.78 |
Specs
| Field | Claude Sonnet 5 | GPT-5.6 Terra |
|---|---|---|
| Publisher | Anthropic | OpenAI |
| Release date | 2026-06-30 | 2026-06-26 |
| Model type | Multimodal model | Reasoning model |
| Architecture | Dense | Dense |
| Parameters | Not available | Not available |
| Context length | 1M | 1.05M |
| Max output | 128K | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | Claude Sonnet 5 | GPT-5.6 Terra |
|---|---|---|
| Text input | $2 / 1M tokens | $2.5 / 1M tokens |
| Text output | $10 / 1M tokens | $15 / 1M tokens |
| Cache read | $0.3 / 1M tokens | $0.25 / 1M tokens |
| Cache write | $3.75 / 1M tokens | $3.125 / 1M tokens |
Summary
- GPT-5.6 Terraleads in:Coding and Software Engineer (2/3), Math and Reasoning (2/2), AI Agent - Tool Usage (1/1), General Evaluation (1/1)
On average across the 7 shared benchmarks, GPT-5.6 Terra scores 10.59 higher.
Largest single-benchmark gap: FrontierMath Tier 4 v2 — Claude Sonnet 5 29.27 vs GPT-5.6 Terra 70.73 (-41.46).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.