AI Rate CardDeepSeek V4 Flash 0731 vs DeepSeek V4 Pro 0813: API price and specs
DeepSeek V4 Flash 0731 is 63% cheaper than DeepSeek V4 Pro 0813 on input price.
DeepSeek V4 Flash 0731 is 93% cheaper than DeepSeek V4 Pro 0813 on output price.
DeepSeek V4 Flash 0731 is 84% cheaper than DeepSeek V4 Pro 0813 on a 1M-in / 200k-out workload.
| Field | DeepSeek V4 Flash 0731 | DeepSeek V4 Pro 0813 |
|---|---|---|
| Lowest input, per 1M | $0.0422 | $0.12 +172.5% |
| Median input, per 1M | $0.14 | $1.32 +836.2% |
| Lowest output, per 1M | $0.09495 | $1.29 +1255.5% |
| Median output, per 1M | $0.30 | $3.96 +1220% |
| Lowest unit price | — | — |
| Cost of 1M in + 200k out | $0.06119 | $0.37 +508.6% |
| Context | 1M | 1M |
| Max output | 1M | 1M |
| Providers | 43 | 37 +14% |
| Regions | CN, EU, GLOBAL, US | APAC, EU, GLOBAL, US |
| Accepts | text | text |
| Returns | text | text |
| Type | Text models | Text models |
| Free option | no | no |
| Cheapest provider | trustedrouter | openrouter |
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
abacus, aihubmix, baseten, cloudflare-workers-ai, cortecs, deepinfra, digitalocean, edenai, empiriolabs, huggingface, hyper, impossibl, io-net, iteracompute, jiekou, kilo, meganova, moark, nebius, novita, ofox, ollama-cloud, openrouter, orcarouter, perplexity-agent, requesty, runinfra, siliconflow, standardcompute, tensorx, togetherai, trustedrouter, vercel, volcengine
Frequently asked questions
Which is cheaper, DeepSeek V4 Flash 0731 or DeepSeek V4 Pro 0813?
DeepSeek V4 Flash 0731 is 63% cheaper than DeepSeek V4 Pro 0813 on input price. DeepSeek V4 Flash 0731 is 93% cheaper than DeepSeek V4 Pro 0813 on output price. DeepSeek V4 Flash 0731 is 84% cheaper than DeepSeek V4 Pro 0813 on a 1M-in / 200k-out workload.
Which has the larger context, DeepSeek V4 Flash 0731 or DeepSeek V4 Pro 0813?
DeepSeek V4 Flash 0731: 1M tokens; DeepSeek V4 Pro 0813: 1M tokens.
Where can I use both DeepSeek V4 Flash 0731 and DeepSeek V4 Pro 0813?
abacus, aihubmix, baseten, cloudflare-workers-ai, cortecs, deepinfra, digitalocean, edenai, empiriolabs, huggingface, hyper, impossibl, io-net, iteracompute, jiekou, kilo, meganova, moark, nebius, novita, ofox, ollama-cloud, openrouter, orcarouter, perplexity-agent, requesty, runinfra, siliconflow, standardcompute, tensorx, togetherai, trustedrouter, vercel, volcengine