DeepSeek

Frontier-adjacent models at a small fraction of Western prices

deepseek.com api-docs.deepseek.com

Overall in llm apis

18of 18 · score 52.4

Data coverage

89%

24 of 27 fields sourced

Best for

•High-volume classification, extraction and summarisation
•Cost-sensitive workloads where a 35x output-price gap against a volume-tier frontier model outweighs a small quality gap
•Self-hosting a capable model under a genuinely permissive licence

Strong on

+Trains on your data Yes 1/18
+$/M cache read $0.0028/M tok 1/12
+Max output 384,000 tokens 1/10
+$/M output $0.28/M tok 3/13
+$/M input $0.14/M tok 3/13

Weak on

−Pinnable versions No 18/18
Better: OpenAI: Yes
−Zero retention No 16/18
Better: Amazon Bedrock: Yes
−Image input No 17/18
Better: Anthropic: Yes
−Batch discount 0% 12/14
Better: Anthropic: 50%

vs Qwen

•Ahead on Max output: 384,000 tokens vs 131,072 tokens
•Ahead on Trains on your data: Yes vs Partial
•Behind on Pinnable versions: No vs Yes
•Behind on Batch discount: 0% vs 50%

How DeepSeek scores among 18 llm apis24 of 27 fields known

Capability7

Flagship model

DeepSeek-V4-Flash (deepseek-v4-flash)

api-docs.deepseek.com · 2026-07-30

Context window1st

1,000,000 tokens

1/14

api-docs.deepseek.com · 2026-07-30

Max output1st

384,000 tokens

1/10

api-docs.deepseek.com · 2026-07-30

Image input

17/18

Leader: Anthropic · Yes

api-docs.deepseek.com · 2026-07-30

Audio in/out

11/18

Leader: OpenAI · Partial

api-docs.deepseek.com · 2026-07-30

Open weights1st

Yes

1/18

github.com · 2026-01-15

OpenAI-compat API1st

Yes

1/18

api-docs.deepseek.com · 2026-07-30

Performance2

TTFT p50

—

Output tok/s

—

Pricing4

$/M input3rd

$0.14/M tok

3/13

Leader: Groq · $0.05/M tok

api-docs.deepseek.com · 2026-07-30

$/M output3rd

$0.28/M tok

3/13

Leader: Fireworks AI · $0.05/M tok

api-docs.deepseek.com · 2026-07-30

$/M cache read1st

$0.0028/M tok

1/12

api-docs.deepseek.com · 2026-07-30

Batch discount

12/14

Leader: Anthropic · 50%

api-docs.deepseek.com · 2026-07-30

Agentic6

Tool use1st

Yes

1/18

api-docs.deepseek.com · 2026-07-30

Schema output

Partial

11/18

Leader: Anthropic · Yes

· 2026-01-15

Effort control

Partial

5/18

Leader: Anthropic · Yes

· 2026-01-15

Computer use

5/18

Leader: Anthropic · Yes

· 2026-01-15

MCP support

6/18

Leader: Anthropic · Yes

· 2026-01-15

Cache TTL

Automatic disk cache, no fixed TTL

api-docs.deepseek.com · 2026-07-30

Governance & continuity6

Continuity policy

The worst on this page for a closed endpoint, and the best if you self-host. The API offers only rolling aliases, deepseek-chat silently points at whatever the current model is, so behaviour can change with no version to pin and no deprecation notice. The mitigation is the MIT licence: download the checkpoint and the version is yours permanently.

· 2026-01-15

Notice period

—

Pinnable versions

18/18

Leader: OpenAI · Yes

· 2026-01-15

Zero retention

16/18

Leader: Amazon Bedrock · Yes

· 2026-01-15

Trains on your data1st