Claude Sonnet 5 vs GPT-5.6 Terra
Claude Sonnet 5 and GPT-5.6 Terra are the production mid-tiers of their labs, priced nearly alike: $2/$10 versus $2/$12 per million tokens. Terra leads the Intelligence Index 57 to 55 and doubles Sonnet's output speed at 149 versus 74 tokens per second. Both carry 1M-token context. Terra wins the measured matchup.
Side by side
| Claude Sonnet 5 | GPT-5.6 Terra | |
|---|---|---|
| Maker | Anthropic | OpenAI |
| Released | 2026-06-30 | 2026-07-09 |
| Context window | 1M tokens | 1M tokens |
| Modalities in | text, image | text, image |
| Input $/MTok | $2 | $2 |
| Output $/MTok | $10 | $12 |
| Open weights | No | No |
Benchmark deltas
| Benchmark | Claude Sonnet 5 | GPT-5.6 Terra | Δ |
|---|---|---|---|
| AA Intelligence Index | 55 | 57 | -2 |
Δ is Claude Sonnet 5 minus GPT-5.6 Terra, in points. Shaded cell wins the row. Dashed values are vendor-published, not yet reproduced — methodology.
Cost calculator
- Claude Sonnet 5
- $200/mo
- GPT-5.6 Terra
- $220/mo
- Difference
- $20/mo
List prices as of 2026-08-11; excludes caching, batch, and long-context tiers — see model pages for tier notes.
Verdict
On the August 2026 numbers, GPT-5.6 Terra is the stronger mid-tier: two Intelligence Index points ahead, twice the output speed, and only $2 more per million output tokens. Sonnet 5's counterargument is qualitative — the Claude line's agentic reliability reputation and ecosystem — which benchmarks compress poorly. Speed-sensitive products and mixed workloads should default to Terra; teams standardized on Claude tooling lose little staying with Sonnet 5 at slightly lower output cost.
Verdicts are analysis, not measurement, and are labeled as such per our editorial policy.
FAQ
- Is Claude Sonnet 5 cheaper than GPT-5.6 Terra?
- It depends on your token mix. Claude Sonnet 5 bills $2 input / $10 output per million tokens; GPT-5.6 Terra bills $2 / $12 (as of August 11, 2026). Output-heavy workloads change the ranking.
- Which has the longer context window — Claude Sonnet 5 or GPT-5.6 Terra?
- They match: both accept 1M tokens of context.
- Which is the more capable model, Claude Sonnet 5 or GPT-5.6 Terra?
- GPT-5.6 Terra scores higher on the Artificial Analysis Intelligence Index: 57 (measured August 11, 2026) against 55 for Claude Sonnet 5 (measured August 11, 2026).
Comparison published August 11, 2026; underlying data last changed August 11, 2026.