DeepSeek · Active

deepseek-flash

deepseek-flash
Context window
1,000,000 tokens
Maximum output
384,000 tokens
Input price
$0.3 per 1M tokens
Output price
$1.2 per 1M tokens
Modalities
text, image

Last verified · Source ↗

Official description

The official model table lists support for thinking and non-thinking requests, structured JSON output, and tool calls.

Read the official model reference ↗.

Where this model fits

deepseek-flash is part of the DeepSeek API catalog. Its documented modalities are text, image. Choose it against the request format, output requirements and operational limits of your application.

The source metadata identifies support for coding. Confirm endpoint-specific parameters in the official reference before integrating that capability.

The source metadata identifies support for tools. Confirm endpoint-specific parameters in the official reference before integrating that capability.

The source metadata identifies support for reasoning. Confirm endpoint-specific parameters in the official reference before integrating that capability.

The source metadata identifies support for vision. Confirm endpoint-specific parameters in the official reference before integrating that capability.

For a repeatable evaluation, use representative requests, inspect structured output or tool calls, and log latency and usage in your own account. Catalog metadata establishes compatibility; it does not measure answer quality on your workload. Start from the provider guide for authentication and endpoint setup.

Current pricing

Verified model prices
ModelPrice typeUSDUnitTierRegionSource
deepseek-flashCached Input$0.003per 1M tokensoff-peakSee sourceOfficial source ↗
deepseek-flashCached Input$0.006per 1M tokenspeakSee sourceOfficial source ↗
deepseek-flashInput$0.15per 1M tokensoff-peakSee sourceOfficial source ↗
deepseek-flashInput$0.3per 1M tokenspeakSee sourceOfficial source ↗
deepseek-flashOutput$0.6per 1M tokensoff-peakSee sourceOfficial source ↗
deepseek-flashOutput$1.2per 1M tokenspeakSee sourceOfficial source ↗

Last verified · Source ↗

Estimate request, daily and monthly usage in the API cost calculator.

Rate limits and availability

Verified rate limits
Model or scopeTierMetricLimitNotesSource
deepseek-flash
deepseek-flash		account defaultconcurrency2,500Account-wide active requests per model; not a requests-per-minute quota.Official source ↗

Last verified · Source ↗

Account-specific capacity may differ. Check the provider rate-limit guide and the free-tier conditions.

Three alternatives to compare

These alternatives share a documented modality and have the closest combination of input price and context size in the current catalog. This is a specification comparison, not a quality benchmark.

Computed model alternatives
ModelProviderInput / 1M tokensContext
Amazon: Nova 2 LiteOpenRouter$0.31,000,000
Qwen: Qwen3.5 Plus 2026-04-20OpenRouter$0.31,000,000
Gemini 2.5 FlashGoogle$0.31,048,576

Use the exact identifier in Python

This identifier fragment belongs in the authenticated request described in the provider Python tutorial. The tutorial covers the correct SDK, endpoint, environment variable and response shape.

model = "deepseek-flash"
# Pass model to the provider-specific request in the tutorial.

Recent changes

  1. DeepSeek · deepseek-flash — MetadataModalities: ["text"] → Modalities: ["text","image"]Source ↗
  2. DeepSeek · deepseek-flash — ValueNot previously recorded → Tier: account default · Metric: concurrency · Value: 2500 · Notes: Account-wide active requests per model; not a requests-per-minute quota.Source ↗
  3. DeepSeek · deepseek-flash — Amount UsdNot previously recorded → Price category: output · USD: 1.2 · Unit: per 1M tokens · Tier: peakSource ↗
  4. DeepSeek · deepseek-flash — Amount UsdNot previously recorded → Price category: output · USD: 0.6 · Unit: per 1M tokens · Tier: off-peakSource ↗
  5. DeepSeek · deepseek-flash — Amount UsdNot previously recorded → Price category: input · USD: 0.3 · Unit: per 1M tokens · Tier: peakSource ↗
  6. DeepSeek · deepseek-flash — Amount UsdNot previously recorded → Price category: input · USD: 0.15 · Unit: per 1M tokens · Tier: off-peakSource ↗
  7. DeepSeek · deepseek-flash — Amount UsdNot previously recorded → Price category: cached_input · USD: 0.006 · Unit: per 1M tokens · Tier: peakSource ↗
  8. DeepSeek · deepseek-flash — Amount UsdNot previously recorded → Price category: cached_input · USD: 0.003 · Unit: per 1M tokens · Tier: off-peakSource ↗
  9. DeepSeek · deepseek-flash — MetadataRecord updated; consult the linked source for details. → Context tokens: 1000000 · Maximum output tokens: 384000Source ↗
  10. DeepSeek · deepseek-flash — First SeenNot previously recorded → Model: deepseek-flashSource ↗

Subscribe to the changelog RSS feed

Frequently asked questions

What is the exact API model identifier?
Use deepseek-flash exactly as shown in the provider model reference. Display names are not interchangeable with API identifiers.
How much does this model cost?
The current verified prices and billing units are listed in the pricing table above. Cache, batch, tier and region conditions remain separate rows when the provider documents them.
How much input can the model accept?
The model card shows the documented context window and maximum output when available. Your request must leave room for both input and generated output, including any provider-specific reasoning allocation.
Can I use a free tier?
Check the linked provider free-tier page for eligibility and account conditions. An account credit or temporary trial does not establish a permanent free rate.
Where do these model facts come from?
Official provider references are linked beside the data and in Sources. Changes are recorded with observation times; unresolved or undocumented fields are left unavailable.

Sources

Last verified · Source ↗