DeepSeek / portfolio
DeepSeek family
DeepSeek checkpoint names and hosted aliases can diverge. Current direct-API documents conflict about the Pro alias; the catalogue rows below retain their accepted identities.
Family research reviewed · Official source notes below
- Standard models
- 15
- Independently rated
- 10 / 15
- Known token rates
- 15 / 15
- Global balanced frontier
- 1
Before you choose
Unresolved direct-API alias
The dated routing announcement says the Pro alias redirects to V4.1-Flash, while the pricing table still identifies V4-Pro-0813. No inference request was made to resolve the discrepancy. Do not transfer ratings across those identities.
Distilled checkpoints
R1 Distill-Llama and Qwen-based distill checkpoints are distinct from full R1 and retain base-model licence considerations.
Compare exact catalogue rows
USD per 1M tokens. In / out are listed base rates; expand billing conditions before planning a request. LMArena scores measure the general preference board, not success on your engineering task. Unknown stays unknown.
V4.1
| Model | LMArena | Input / output $/M | Context / max output | Capabilities & conditions |
|---|---|---|---|---|
DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash | Unrated | $0.100 / $0.500Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 1,048,576Output 943,718 | text · imageTools yes · reasoning yesWeights unknownReleased 2026-09-10 |
V4 dated and earlier checkpoints
| Model | LMArena | Input / output $/M | Context / max output | Capabilities & conditions |
|---|---|---|---|---|
DeepSeek V4 Flash Vision Expdeepseek/deepseek-v4-flash-vision-exp | Unrated | $0.220 / $0.660Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 1,048,576Output 943,718 | text · imageTools yes · reasoning yesClosed weightsReleased 2026-08-21 |
DeepSeek V4 Pro 0813deepseek/deepseek-v4-pro-0813 | Unrated | $0.462 / $1.39Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 1,048,576Output 384,000 | textTools yes · reasoning yesWeights unknownReleased 2026-08-12 |
DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731 | Unrated | $0.040 / $0.640Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 1,310,720Output 943,718 | textTools yes · reasoning yesWeights unknownReleased 2026-07-31 |
DeepSeek V4 Pro 0423deepseek/deepseek-v4-pro | 1444.0±6.8 interval half-widthEffort: high | $0.946 / $1.89Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 1,048,576Output 384,000 | textTools yes · reasoning yesOpen weightsReleased 2026-04-24 |
DeepSeek V4 Flash 0423deepseek/deepseek-v4-flash | 1431.8±4.1 interval half-widthEffort: default | $0.083 / $0.165Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 1,048,576Output 384,000 | textTools yes · reasoning yesOpen weightsReleased 2026-04-24 |
R1 distill checkpoint
| Model | LMArena | Input / output $/M | Context / max output | Capabilities & conditions |
|---|---|---|---|---|
R1 Distill Llama 70Bdeepseek/deepseek-r1-distill-llama-70b | Unrated | $0.800 / $0.800Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 8,192Output 7,372 | textTools no · reasoning yesWeights unknown · lifecycle date 2026-09-28Released 2025-01-23 |
R1
| Model | LMArena | Input / output $/M | Context / max output | Capabilities & conditions |
|---|---|---|---|---|
R1 0528deepseek/deepseek-r1-0528 | 1427.5±5.7 interval half-widthEffort: default | $0.500 / $2.15Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 163,840Output 32,768 | textTools yes · reasoning yesWeights unknownReleased 2025-05-28 |
R1deepseek/deepseek-r1 | 1372.6±4.8 interval half-widthEffort: default | $0.700 / $2.50Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 64,000Output 16,000 | textTools yes · reasoning yesWeights unknownReleased 2025-01-20 |
V3 and earlier chat
| Model | LMArena | Input / output $/M | Context / max output | Capabilities & conditions |
|---|---|---|---|---|
DeepSeek V3.2deepseek/deepseek-v3.2 | 1424.8±3.5 interval half-widthEffort: default | $0.269 / $0.400Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 163,840Output 65,536 | textTools yes · reasoning yesOpen weights · lifecycle date 2026-09-28Released 2025-12-01 |
DeepSeek V3.2 Expdeepseek/deepseek-v3.2-exp | 1422.6±6.4 interval half-widthEffort: default | $0.270 / $0.410Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 163,840Output 65,536 | textTools yes · reasoning yesWeights unknown · lifecycle date 2026-09-28Released 2025-09-29 |
DeepSeek V3.1 Terminusdeepseek/deepseek-v3.1-terminus | 1416.9±9.7 interval half-widthEffort: default | $0.270 / $1.00Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 163,840Output 32,768 | textTools yes · reasoning yesWeights unknown · lifecycle date 2026-09-28Released 2025-09-22 |
DeepSeek V3.1deepseek/deepseek-chat-v3.1 | 1419.7±6.0 interval half-widthEffort: default | $0.250 / $0.950Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 163,840Output 32,768 | textTools yes · reasoning yesWeights unknownReleased 2025-08-21 |
DeepSeek V3 0324deepseek/deepseek-chat-v3-0324 | 1375.0±3.9 interval half-widthEffort: default | $0.250 / $1.00Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 163,840Output 147,456 | textTools yes · reasoning noWeights unknownReleased 2025-03-24 |
DeepSeek V3deepseek/deepseek-chat | 1332.4±4.7 interval half-widthEffort: default | $0.320 / $0.890Listed base ratesBilling details
Tool, media, cache-write and service-tier charges may add dimensions. Read the full model evidence. | 163,840Output 16,384 | textTools yes · reasoning noWeights unknownReleased 2024-12-26 |
Free and batch variants
These rows have different access or processing conditions. They are excluded from the standard-row counts above. A variant without a standard base remains a separately named variant.
- DeepSeek V4.1 Flash (batch)batch · $0.112 in / $0.336 out per 1M
Named, but absent from the accepted catalogue
deepseek/deepseek-v4-flash-vision-exp:batchhas no accepted row. No price or score is inferred.deepseek/deepseek-v4-pro-0813:batchhas no accepted row. No price or score is inferred.deepseek/deepseek-v4-flash-0731:batchhas no accepted row. No price or score is inferred.
Sources and scope
- DeepSeek V4.1 Flash announcement ↗
DeepSeek first-party API aliases only
Checked 2026-09-22 · web timed out; official URL fetched directly with HTTP 200 and inspected - DeepSeek model details ↗
First-party API table; unresolved documentation conflict
Checked 2026-09-22 · web timed out; official URL fetched directly with HTTP 200 and inspected - Official DeepSeek V4.1 Flash card ↗
Exact named models on the vendor platform; not a promise about every reseller
Checked 2026-09-22 · official document opened and inspected - Official DeepSeek R1 card ↗
Exact named models on the vendor platform; not a promise about every reseller
Checked 2026-09-22 · official document opened and inspected
Source review does not refresh a price or transplant a benchmark score. Read the method and correction policy.
Questions this page answers
Are the DeepSeek family models interchangeable?
No. These are exact catalogue rows with separate capability, pricing and rating evidence. Compare the checkpoint, serving endpoint, reasoning settings and required input types. A family name does not establish equivalent behaviour.
How many DeepSeek rows are compared here?
15 standard rows, of which 10 have an accepted LMArena general score and 15 have known input and output token rates. 1 related free or batch variants are listed separately.
What do the prices and frontier counts mean?
Prices below are listed US dollars per million input or output tokens, before request-specific context tiers and other billing dimensions. The frontier count comes from the global balanced view. Neither a family price range nor a benchmark difference proves your task will be cheaper or more successful.
How current are the official notes?
The linked family research was reviewed 2026-09-22. That date is separate from each accepted catalogue observation and each vendor’s effective date. An official note applies to its named platform, not automatically to every reseller.