Select exact model records. See observed capabilities together, then inspect benchmark results that share a source, version and recorded methodology.
| Source field | GPT 6.1 Sol ↗ | Claude Sonnet 5.5 ↗ |
|---|---|---|
| Publisher | OpenAI | Anthropic |
| Access | Hosted model | Hosted model |
| Input types | text | text, image |
| Output types | text | text |
| Context window | Not established | 1,000,000 |
| Output allowance | Not established | 128,000 |
| Original release date | Sep 29, 2026 | Sep 28, 2026 |
| Verification | verified | verified |
| Record last observed | Oct 10, 2026 | Oct 13, 2026 |
| Benchmark / version | GPT 6.1 Sol | Claude Sonnet 5.5 |
|---|---|---|
| Artificial Analysis Intelligence Indexv4.3 · Artificial Analysis | 51.8 index_pointsSep 30, 2026Measurement source ↗ | 56 index_pointsSep 29, 2026Measurement source ↗ |
Enter results from your own evaluation. A task should count once it meets your written quality criteria. Include failed attempts and retries in total spend. These inputs stay in this browser and are not a Flonno benchmark.
Enter all four values. Attempted tasks must be at least one; accepted tasks cannot exceed attempted tasks.
Enter all four values. Attempted tasks must be at least one; accepted tasks cannot exceed attempted tasks.
A shared benchmark method can make two reported scores comparable, but it does not establish quality in your application. Test representative inputs, full request latency, output quality and cost per accepted result on your selected deployment.
Quantized derivatives and different inference settings need their own measurements. Read the evidence methodology or estimate your token budget.
