Mixed: deepseek/deepseek-v4-flash on relace
| Field | Before | After | Change |
|---|---|---|---|
| Input | 0.0228 | 0.0500 | +119.3% |
| Output | 1.28 | 0.45 | -64.8% |
| Cache read | 0.0228 | 0.0500 | +119.3% |
What this means by workload
Illustrative monthly cost of the changed field(s) alone, at a representative 10M tokens/month split per workload preset. Not a substitute for the calculator, which uses your own volume and every price field.
| Workload | Before / month | After / month |
|---|---|---|
| Coding agent | $0.86 | $0.70 |
| RAG chat | $2.11 | $1.10 |
| Summarisation (batch) | $1.49 | $0.90 |
| Classification | $0.48 | $0.58 |
| Chat assistant | $5.26 | $2.10 |
| Content generation | $10.29 | $3.70 |
Back to model page | Compare vs Llama 3.3 70B Instruct | All changes