AI Rate CardDeepSeek V3.2 vs Phi-4: API price and specs
Phi-4 is 75% cheaper than DeepSeek V3.2 on input price.
Phi-4 is 66% cheaper than DeepSeek V3.2 on output price.
Phi-4 is 73% cheaper than DeepSeek V3.2 on a 1M-in / 200k-out workload.
DeepSeek V3.2 has a 28.1% larger context than Phi-4.
| Field | DeepSeek V3.2 | Phi-4 |
|---|---|---|
| Lowest input, per 1M | $0.22 +292.9% | $0.056 |
| Median input, per 1M | $0.30 +328.6% | $0.07 |
| Lowest output, per 1M | $0.33 +194.6% | $0.11 |
| Median output, per 1M | $0.50 +257.1% | $0.14 |
| Lowest unit price | — | — |
| Cost of 1M in + 200k out | $0.29 +264.8% | $0.0784 |
| Context | 164K | 128K +22% |
| Max output | 164K | 15K +91% |
| Providers | 29 | 10 +65.5% |
| Regions | EU, GLOBAL, US | US |
| Accepts | text | text |
| Returns | text | text |
| Type | Text models | Text models |
| Free option | no | no |
| Cheapest provider | tokengo | kilo |
Your own workload: how many tokens the model gets (input) and gives back (output) - for example, in a month. The "Cost" row of the table above is recalculated for these numbers, for each model.
Computed from current offers each time the page opens; the best value of a row is bold, the percent is how far behind the best.
Providers selling both
azure, deepinfra, edenai, fastrouter, huggingface, impossibl, kilo, openrouter, requesty, trustedrouter
Frequently asked questions
Which is cheaper, DeepSeek V3.2 or Phi-4?
Phi-4 is 75% cheaper than DeepSeek V3.2 on input price. Phi-4 is 66% cheaper than DeepSeek V3.2 on output price. Phi-4 is 73% cheaper than DeepSeek V3.2 on a 1M-in / 200k-out workload.
Which has the larger context, DeepSeek V3.2 or Phi-4?
DeepSeek V3.2: 164K tokens; Phi-4: 128K tokens.
Where can I use both DeepSeek V3.2 and Phi-4?
azure, deepinfra, edenai, fastrouter, huggingface, impossibl, kilo, openrouter, requesty, trustedrouter