deepseek/deepseek-v4-flash
estimated usage cost
$0.445
1,000 requests × (1,000 uncached input × $0.19 + 500 output × $0.51) / 1,000,000 = $0.445
base rates; no context pricing tiers are published.
DeepSeek V4 Flash vs Phi-4
rates, context and capabilities side by side on the impossibl api, and what the same workload costs on each.
| specification | DeepSeek V4 Flash deepseek/deepseek-v4-flash | Phi-4 microsoft/phi-4 |
|---|---|---|
| input · usd / 1m tok | $0.19 | $0.125 |
| output · usd / 1m tok | $0.51 | $0.50 |
| context window | 1mtokens | 16ktokens |
| modalities · in → out | takes text in, gives text out | takes text in, gives text out |
pricing
| base api rates · usd per 1m tokens | DeepSeek V4 Flash deepseek/deepseek-v4-flash | Phi-4 microsoft/phi-4 |
|---|---|---|
| input | $0.19 | $0.125 |
| output | $0.51 | $0.50 |
| cache read | $0.028 | not published |
the estimate below prices one workload at these rates on every model.
specifications
| as published | DeepSeek V4 Flash deepseek/deepseek-v4-flash | Phi-4 microsoft/phi-4 |
|---|---|---|
| creator | DeepSeek | Microsoft |
| model type | Text generation | Text generation |
| context window | 1,000,000tokens | 16,000tokens |
| input | takes text in, gives text out | takes text in, gives text out |
| output | text | text |
| tool calling | not verified | not supported |
| released | apr 24, 2026 | dec 12, 2024 |
| endpoint | /v1/chat/completions | /v1/chat/completions |
unknown means the public catalog does not establish that fact. a model's documented capabilities can differ from what a particular api supports. see the model documentation for integration details.
benchmarks
| primary configuration | DeepSeek V4 Flash deepseek/deepseek-v4-flash max | Phi-4 microsoft/phi-4 |
|---|---|---|
| intelligence index · gateway rank | 34.3#44 of 114 | 5.9#113 of 114 |
| coding index | 69.1#25 of 66 | not measured |
| agentic index | 41.0#21 of 54 | not measured |
| omniscience | -14.3#60 of 90 | -55.7#87 of 90 |
| gpqa | 90.8%#22 of 82 | 57.5%#76 of 82 |
| humanity’s last exam | 38.6%#33 of 93 | 3.8%#88 of 93 |
| aa-lcr | 79.7% | 0.0% |
| automationbench partial | 54.0% | not measured |
| scicode | 50.3% | not measured |
| tau banking | 39.4% | not measured |
| critpt | 16.6% | 0.0% |
| gdp.pdf all-pass | 11.0% | not measured |
| ifbench | not measured | 23.5% |
| livecodebench | not measured | 23.1% |
| aime 2025 | not measured | 18.0% |
| terminal-bench hard | not measured | 3.8% |
| tau2 | not measured | 0.0% |
DeepSeek V4 Flash source ↗Phi-4 source ↗
each column describes the tested configuration named under the model; performance and costs can differ on impossibl. a score one model was not measured on is left blank rather than scored zero. indices are ranked against every model on the gateway with one. last checked 2026-10-27.
estimate your cost
one token workload, priced on every model at the published rates, in usd. change the numbers to match yours.
microsoft/phi-4
estimated usage cost
$0.375
1,000 requests × (1,000 uncached input × $0.125 + 500 output × $0.50) / 1,000,000 = $0.375
base rates; no context pricing tiers are published.
estimates use published impossibl api usage rates. configured billing adjustments, funding fees, taxes, and custom workspace rates are excluded and may change the final amount.
image and audio usage, cache creation, and additional tool charges are excluded. reasoning tokens count as billable output, even when they are not visible in the response.
summary
DeepSeek V4 Flash and Phi-4 are available through the Impossibl API. they share the /v1/chat/completions endpoint; the model identifier selects the model for a request.
DeepSeek V4 Flash is created by DeepSeek. It has a 1,000,000-token context window. Base API pricing (USD): $0.19 per million input tokens; $0.51 per million output tokens.
Phi-4 is created by Microsoft. It has a 16,000-token context window. Base API pricing (USD): $0.125 per million input tokens; $0.50 per million output tokens.
this page compares DeepSeek V4 Flash vs Phi-4 as listed by the impossibl public api. prices and availability come from that catalog, refreshed here every five minutes. this is an impossibl offering, not a comparison of independent hosting-provider quotes.
view the source catalog ↗comparison as markdown ↗report a correction →