claude-sonnet-5-5 is Anthropic's production-tier model in the Claude 5 family, priced at $2.00 per million input tokens and $10.00 per million output tokens. Prompt caching reduces repeated input to $0.10/1M. It supports a 1M-token context window, vision inputs, function calling, and streaming. claude-sonnet-5-5 is a proprietary model available exclusively through Anthropic's API.
claude-sonnet-5-5 is priced at $2.00 per million input tokens and $10.00 per million output tokens, with prompt caching at $0.10/1M. At 1,000 input + 500 output tokens per call, the cost is ($2.00 × 1,000 + $10.00 × 500) ÷ 1,000,000 = $0.007 per call, or approximately $7.00 per 1,000 API calls. The 5:1 output-to-input ratio is consistent with Anthropic's Claude 5 lineup. Prompt caching at $0.10/1M is among the lowest cached input rates in the Claude 5 family, making it particularly cost-effective for workloads with long repeated system prompts.
claude-sonnet-5-5's 1M-token context window enables large-document and codebase analysis in a single API call. Median latency is approximately 700ms TTFT (Estimate). Throughput is approximately 95 tokens per second (Estimate). Maximum output is 32,000 tokens. As a standard (non-reasoning) model, it delivers faster responses than the reasoning-tier Claude 5 models, making it suitable for interactive and latency-sensitive applications.
claude-sonnet-5-5 is best suited to production workloads requiring a balance of capability, cost, and speed: customer-facing chat assistants, document analysis, code generation and review, vision-based workflows, and agentic pipelines. Its 1M-token context window and competitive pricing make it a strong default for most Claude 5 deployments. For tasks requiring extended reasoning, claude-opus-5-5 or the frontier-tier models offer stronger performance at higher cost.
claude-sonnet-5-5 sits at the production tier of Anthropic's Claude 5 lineup, positioned between the efficient tier (claude-haiku-5-5 at $0.10/$0.50) and the advanced reasoning tier (claude-opus-5-5 at $4/$20). Its intelligence score of approximately 91/100 reflects strong production-tier capability. Main competitors are production-tier models from OpenAI and Google at similar price points. For most Claude 5 deployments, claude-sonnet-5-5 is the recommended starting point — upgrade to reasoning models only when task complexity justifies the cost premium.
Composite quality score based on reasoning, coding, and instruction-following benchmarks.
Input price vs 135 frontier models tracked. Avg: $5.69/1M.
| Model ID | claude-sonnet-5-5 |
| Provider | Anthropic |
| Model Family | Claude 5 |
| Tier | Frontier |
| Context Window | 1,000,000 tokens (1.0M) |
| Max Output Tokens | 32,000 tokens |
| Input Price | $2 / 1M tokens |
| Output Price | $10 / 1M tokens |
| Cached Input Price | $0.1 / 1M tokens |
| Throughput | 95 tokens/sec |
| Latency (P50) | 700ms |
| Intelligence Score | 91/100 |
| LiveBench Score | — |
| Cost per Task (LiveBench) | — |
| Data Confidence | Partial |
| HLE Score | — |
| GPQA Diamond | — |
| SWE-bench Score | — |
| OSWorld Score | — |
| BrowseComp Score | — |
| Terminal-Bench Score | — |
| AutoBench Score | — |
| Parameters | — |
| Vision Support | Yes |
| Function Calling | Yes |
| Streaming | Yes |
| Reasoning Model | No |
| Open Weights | No |
| Data Source | Static (verified) |
| Last Updated | 2026-10-23 |
Similar Models by Intelligence Score
Self-hosting claude-sonnet-5-5? These GPUs are well-matched for this model's compute and memory requirements.