Claude Sonnet 5 vs GPT-5.5
Pricing, benchmarks, and use case comparison
Newer versions are out: Claude Sonnet 5 has been replaced by Claude Sonnet 5.5; GPT-5.5 has been replaced by GPT-6.1 Sol.
Verdict
Our pick: Claude Sonnet 5Pick Claude Sonnet 5: at $2/$10 it costs far less than GPT-5.5 at $5/$30 and posts the higher SWE-bench Pro score. Both are previous-generation models now; for new work compare Claude Sonnet 5.5 with GPT-6.1 Sol, which both list at $2/$10.
Specs comparison
| Claude Sonnet 5 | GPT-5.5 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Type | Closed source | Closed source |
| Context window | ✓1M | 1,050,000 tokens (128,000 max output) |
| Input / 1M tokens | ✓$2.00 | $5.00 |
| Output / 1M tokens | $10.00 | $30.00 |
| Release date | 2026-06 | 2026-04 |
Benchmarks
| Benchmark | Claude Sonnet 5 | GPT-5.5 |
|---|---|---|
| SWE-bench Verified | ✓~85.2% | 82.6% |
| Terminal-Bench 2.1 | 80.4% | - |
| SWE-bench Pro | ✓63.2% | 58.6% |
| OSWorld-Verified | 81.2% | - |
Scores sourced from official provider release posts and independent benchmark aggregators.
Both have successors: Claude Sonnet 5 was followed by Claude Sonnet 5.5, and OpenAI's current lineup above GPT-5.5 is the GPT-6 family, led by GPT-6 Astra and GPT-6.1 Sol.
Capability and benchmarks
GPT-5.5 posts 82.6% SWE-bench Verified and 58.6% SWE-bench Pro. Sonnet 5 reports ~85.2% SWE-bench Verified (secondary sources), a higher 63.2% SWE-bench Pro, plus 80.4% Terminal-Bench 2.1 and 81.2% OSWorld-Verified. On our capability index they are close (Sonnet coding 92 / reasoning 90 vs GPT-5.5 coding 93 / reasoning 92), so price and context matter more than raw ability.
Price and context
Sonnet 5 is much cheaper: $2/$10 per 1M, now its permanent price, against GPT-5.5's $5/$30. GPT-5.5 offers a slightly larger context (1,050,000 tokens vs Sonnet's 1M) but bills prompts above 272K input tokens at 2x input and 1.5x output, while Sonnet 5 charges the standard rate across its full window. Both accept text and image input. Sonnet 5 is also faster (capability speed 82 vs 62), though its tokenizer produces about 30% more tokens for the same text than Sonnet 4.6. Both remain available: OpenAI has announced no deprecation for GPT-5.5, and Anthropic lists Sonnet 5 as a legacy model with retirement not sooner than June 30, 2027.
Which to pick
- Pick Sonnet 5 for cost-sensitive, high-volume agent loops, long prompts, and terminal-agent and computer-use work.
- Pick GPT-5.5 only to keep an existing pipeline tuned to its behavior; GPT-6.1 Sol now costs $2/$10 and is OpenAI's cheaper, newer option (see Claude Sonnet 5.5 vs GPT-6.1 Sol).
Which should you choose?
Choose Claude Sonnet 5 if...
- →You have a pipeline tuned to Sonnet 5 and want stable behavior while you test Sonnet 5.5
- →You turn thinking off with thinking type disabled or use forced tool_choice, both of which Sonnet 5.5 rejects
- →You need a 1M context window at $2/$10 with the Sonnet 5 request surface
Choose GPT-5.5 if...
- →You have prompts and evals tuned to GPT-5.5 and want stable behavior
- →Complex professional coding, data analysis, and multi-tool agentic workflows on an existing GPT-5.5 integration
- →Long-document or large-codebase tasks needing a 1M+ context window