API price comparison · Updated
Claude Sonnet 5.5 costs $2.00 per million input tokens and $10.00 per million output tokens; DeepSeek V4.1 Flash costs $0.30 and $1.20. For an agent task (80,000 input tokens, 60,000 of them cached, and 6,000 output), Claude Sonnet 5.5 costs $0.106 and DeepSeek V4.1 Flash $0.014: Claude Sonnet 5.5 costs 7.8× as much.
DeepSeek V4.1 Flash costs 86% less for this workload. Excludes cache writes, tools and taxes.
| Price | Claude Sonnet 5.5 | DeepSeek V4.1 Flash |
|---|---|---|
| Model ID | claude-sonnet-5-5 | deepseek-ai/DeepSeek-V4.1-Flash |
| Input | $2.00 | $0.30 |
| Cache read | $0.10 | $0.006 |
| Cache write | $2.50 | — |
| Output | $10.00 | $1.20 |
| Batch in / out | $1.00 / $5.00 | No Batch rate |
| Long prompts | Same rate | Same rate |
| Scheduled change | None announced | None announced |
| Status | Available | Available |
Prices for each provider’s own API, from Anthropic’s pricing page and Together AI’s pricing page. The lower price in each row is highlighted. Ganivra uses these same rates to price Claude Sonnet 5.5 and DeepSeek V4.1 Flash calls.
Worked out from the prices above at standard rates, with no Batch discount. Long-prompt rates apply where a model has them.
| Request | Claude Sonnet 5.5 | DeepSeek V4.1 Flash | Per 1,000 requests | Difference |
|---|---|---|---|---|
| Chat reply3,000 input tokens, 500 output | $0.011 | $0.0015 | $11.00 / $1.50 | Claude Sonnet 5.5 costs 7.3× as much |
| Document summary30,000 input tokens, 1,500 output | $0.075 | $0.011 | $75.00 / $10.80 | Claude Sonnet 5.5 costs 6.9× as much |
| Agent task80,000 input tokens (60,000 cached), 6,000 output | $0.106 | $0.014 | $106 / $13.56 | Claude Sonnet 5.5 costs 7.8× as much |
| Long document300,000 input tokens, 2,000 output | $0.620 | $0.092 | $620 / $92.40 | Claude Sonnet 5.5 costs 6.7× as much |
Ganivra prices every Claude Sonnet 5.5 and DeepSeek V4.1 Flash call you make and adds it up by customer, feature and workflow, next to what each customer pays. Switch a workflow and see what it did to each customer’s margin.
Per million tokens, Claude Sonnet 5.5 costs $2.00 input and $10.00 output, and DeepSeek V4.1 Flash $0.30 input and $1.20 output. For an agent task (80,000 input tokens, 60,000 of them cached, and 6,000 output), Claude Sonnet 5.5 costs 7.8× as much: $106 against $13.56 per 1,000 tasks.
$2.00 input and $10.00 output per million tokens on Anthropic's API, with cached input at $0.10 and Batch at $1.00 input and $5.00 output. The rate is the same at any prompt length.
$0.30 input and $1.20 output per million tokens on Together AI's API, with cached input at $0.006. The rate is the same at any prompt length.
For a 300,000 input tokens, 2,000 output request, Claude Sonnet 5.5 costs $0.620 and DeepSeek V4.1 Flash $0.092: Claude Sonnet 5.5 costs 6.7× as much. On Claude Sonnet 5.5, one rate at any prompt length; on DeepSeek V4.1 Flash, one rate at any prompt length.
Record what each call actually used, then price the same usage on both models. Ganivra does this from your recorded calls: it shows what each customer, feature and workflow would cost on Claude Sonnet 5.5 and on DeepSeek V4.1 Flash, next to what each customer pays you.
Every model on one page: Claude Sonnet 5.5 pricing · DeepSeek V4.1 Flash pricing · all model price comparisons · all supported models.