DeepSeek V4pro
API
deepseek-v4-protext → text
input → output
DeepSeek API
Source for DeepSeek V4 ProDS
DeepSeek Models
Data verified
Compare the current DeepSeek V4 Pro and Flash families with R1, V3.2, and earlier open-weight releases. Model IDs, reasoning settings, scores, and access routes stay separate.
DeepSeek currently exposes two V4 API model IDs: deepseek-v4-pro and deepseek-v4-flash. Both support a 1M-token context window, thinking and non-thinking modes, and open weights. V4 Pro is the stronger tracked family; V4 Flash is the lower-cost choice for high-volume work. This directory tracks 18 variants across 9 model families.
The model name and the API ID are not always the same thing. DeepSeek uses one API ID per V4 family, then switches thinking and reasoning effort through request settings. We keep those configurations as separate benchmark rows because their scores differ, but we do not pretend they are separate billable SKUs.
Older R1, V3.2, V3.1, Coder, and Math rows remain useful for reproducibility and self-hosting. They are not current first-party API choices. DeepSeek set July 24, 2026 as the retirement deadline for the deepseek-chat and deepseek-reasoner aliases, so new integrations should use the two V4 IDs above.
DeepSeek V4 Pro vs V4 Flash
Start with the family, then choose a reasoning setting. The public score below is the strongest tracked configuration in each family, not a promise that every prompt or API mode will reproduce it.
| Family | Current API ID | Strongest tracked configuration | Public score | Context | Use it when |
|---|---|---|---|---|---|
| DeepSeek V4 Pro | deepseek-v4-pro | Not ranked | — | 1M | Coding, agents, math, and harder reasoning justify the extra cost. |
| DeepSeek V4 Flash | deepseek-v4-flash | Not ranked | — | 1M | Extraction, classification, summarization, and throughput matter most. |
Our default: start with V4 Flash, measure task failures, then route only the failing slice to V4 Pro. That keeps the model decision tied to an acceptance test instead of treating a leaderboard score as a universal routing rule.
Honest limit. The directory does not currently have comparable first-party throughput or uptime measurements for both V4 endpoints. If latency or regional availability decides the deployment, run the same workload against the live endpoints before choosing.
DeepSeek model release history
| Release | Date | What changed | Current role |
|---|---|---|---|
| DeepSeek V4 | Apr 24, 2026 | Pro and Flash introduced dedicated IDs and a 1M-token context window. | Current API and open-weight families |
| DeepSeek V3.2 | Dec 1, 2025 | Unified thinking and non-thinking modes in the V3 line. | Historical and self-hosted comparison |
| DeepSeek R1 | Jan 20, 2025 | Released the reasoning model and distilled open-weight variants. | Self-hosting and reasoning-history reference |
| DeepSeek V3 | Dec 26, 2024 | Established the 671B mixture-of-experts V3 family. | Historical and open-weight reference |
DeepSeek models FAQ
Which is the best DeepSeek model?
DeepSeek V4 Pro is the stronger current family in the tracked benchmark rows. Its Max reasoning configuration scores —/100. Use V4 Flash when throughput and token cost matter more, then escalate only the requests that fail your acceptance tests.
What DeepSeek models are available through the API?
The current first-party API documentation lists deepseek-v4-pro and deepseek-v4-flash. Each ID supports thinking and non-thinking modes. The older deepseek-chat and deepseek-reasoner aliases reached their retirement deadline on July 24, 2026.
What is the difference between DeepSeek V4 Pro and Flash?
V4 Pro is the larger family and leads the tracked DeepSeek rows for demanding coding, agentic, and reasoning work. V4 Flash trades some benchmark performance for a lower API price. Both expose a 1M-token context window and configurable thinking modes.
Are DeepSeek models open source?
DeepSeek publishes weights for V4, R1, V3.2, and several earlier families. We label them open weight: downloadable weights are clear evidence, while full training-data and reproducibility openness is a broader claim. Hosted API use still incurs token charges.
First-party DeepSeek sources
Model count
9
Best model
54.43/100
DeepSeek V3.2
Avg. top 3
53.7/100
Latest release
August 2026
Public-ranked
8
Supported
4
Current provider directory
Current DeepSeek models and API IDs
As of August 12, 2026, this directory contains 8 active, preview, announced, or limited-access entries. Provider-issued IDs appear only when the underlying documentation identifies them; BenchLM model keys are kept separate.
Latest tracked release
DeepSeek V4 Pro, DeepSeek V4 Pro (High), and DeepSeek V4 Pro (Max)
Aug 13, 2026
Open the dated sourceOperational record
- Current API IDs
- 2
- Inactive variants
- 0
Current provider-issued API model IDs
Shared IDs are listed once. Request-time reasoning modes and other configurations do not become separate model IDs.
deepseek-v4-proDeepSeek V4 Pro + 2 tracked variants
Source for deepseek-v4-prodeepseek-v4-flashDeepSeek V4 Flash + 2 tracked variants
Source for deepseek-v4-flash
Earlier source-dated releases
Together with the latest release above, these are the three newest dated release groups in the tracked provider inventory.
DeepSeek V4 Flash, DeepSeek V4 Flash (High), and DeepSeek V4 Flash (Max)
Jul 31, 2026
Dated sourceDeepSeek V4 Flash Base and DeepSeek V4 Pro Base
Apr 24, 2026
Dated source
Score trend
DeepSeek's average score across its top 3 ranked models, cumulative through each release month.
All DeepSeek models
Search every tracked model variant, including preview, limited-access, deprecated, and open-weight releases. Provider identifiers and specifications link to first-party evidence where it is available.
- Model variants
- 18
- Model families
- 9
- Current entries
- 8
- Evidence date
- August 12, 2026
Showing 18 of 18 variants
Prices are input / output per 1M tokens.
- Model
- DeepSeek V4 Pro
- Lifecycle
- Available
- Official identifier
- Modalities
- Context
- 1M
- Price
- $0.43 / $0.87
- Access and source
- Model
- DeepSeek V4 Pro (High)
DeepSeek V4pro-reasoning- Lifecycle
- Available
- Official identifier
APIdeepseek-v4-proThinking mode with high effort on the shared family API ID.
- Modalities
text → text
input → output
- Context
- 1M
- Price
- $0.43 / $0.87
- Access and source
DeepSeek APISource for DeepSeek V4 Pro (High)- Model
- DeepSeek V4 Pro (Max)
DeepSeek V4- Lifecycle
- Available
- Official identifier
APIdeepseek-v4-proThinking mode with max effort on the shared family API ID.
- Modalities
text → text
input → output
- Context
- 1M
- Price
- $0.43 / $0.87
- Access and source
DeepSeek APISource for DeepSeek V4 Pro (Max)- Model
- DeepSeek V4 Flash
DeepSeek V4flash- Lifecycle
- Public beta
- Official identifier
APIdeepseek-v4-flash- Modalities
text → text
input → output
- Context
- 1M
- Price
- $0.14 / $0.28
- Access and source
DeepSeek APISource for DeepSeek V4 Flash- Model
- DeepSeek V4 Flash (High)
DeepSeek V4flash-reasoning- Lifecycle
- Public beta
- Official identifier
APIdeepseek-v4-flashThinking mode with high effort on the shared family API ID.
- Modalities
text → text
input → output
- Context
- 1M
- Price
- $0.14 / $0.28
- Access and source
DeepSeek APISource for DeepSeek V4 Flash (High)- Model
- DeepSeek V4 Flash (Max)
DeepSeek V4flash-reasoning- Lifecycle
- Public beta
- Official identifier
APIdeepseek-v4-flashThinking mode with max effort on the shared family API ID.
- Modalities
text → text
input → output
- Context
- 1M
- Price
- $0.14 / $0.28
- Access and source
DeepSeek APISource for DeepSeek V4 Flash (Max)- Model
- DeepSeek V4 Flash Base
DeepSeek V4base- Lifecycle
- Available
- Official identifier
Weightsdeepseek-ai/DeepSeek-V4-Flash- Modalities
text → text
input → output
- Context
- 1M
- Price
- Not documented
- Access and source
open weightsSource for DeepSeek V4 Flash Base- Model
- DeepSeek V4 Pro Base
DeepSeek V4base- Lifecycle
- Available
- Official identifier
Weightsdeepseek-ai/DeepSeek-V4-Pro- Modalities
text → text
input → output
- Context
- 1M
- Price
- Not documented
- Access and source
open weightsSource for DeepSeek V4 Pro Base- Model
- DeepSeek V3.2
DeepSeek V3.2- Lifecycle
- Established
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- $0.28 / $0.42
- Access and source
Not documentedSource for DeepSeek V3.2- Model
- DeepSeek V3.2 (Thinking)
DeepSeek V3.2reasoning- Lifecycle
- Established
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- $0.55 / $2.19
- Access and source
Not documentedSource for DeepSeek V3.2 (Thinking)- Model
- DeepSeek V3.1
DeepSeek V3.1- Lifecycle
- Established
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- Self-host
- Access and source
Not documentedSource for DeepSeek V3.1- Model
- DeepSeek V3.1 (Reasoning)
DeepSeek V3.1reasoning- Lifecycle
- Established
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- Self-host
- Access and source
Not documentedSource for DeepSeek V3.1 (Reasoning)- Model
- DeepSeek R1 Distill Qwen 32B
DeepSeek R1 Distill- Lifecycle
- Established
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- Self-host
- Access and source
Not documented- Model
- DeepSeek-R1
DeepSeek-R1- Lifecycle
- Established
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- $0.55 / $2.19
- Access and source
Not documentedSource for DeepSeek-R1- Model
- DeepSeek V3
DeepSeek- Lifecycle
- Established
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- $0.27 / $1.10
- Access and source
Not documentedSource for DeepSeek V3- Model
- DeepSeek Coder 2.0
DeepSeek Coder 2.0- Lifecycle
- Tracked
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- Not documented
- Access and source
Not documented- Model
- DeepSeek LLM 2.0
DeepSeek LLM 2.0- Lifecycle
- Tracked
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- Self-host
- Access and source
Not documented- Model
- DeepSeekMath V2
DeepSeekMath- Lifecycle
- Tracked
- Official identifier
- Not documented
- Modalities
Not documented
input → output
- Context
- 128K
- Price
- Self-host
- Access and source
Not documented
Prices are input / output per 1M tokens.
Ranked family representatives
One canonical entry per model family. Public scores use the same evidence rules as the main leaderboard.
| Context | Price | Public score | |||||
|---|---|---|---|---|---|---|---|
| DeepSeek V3.2 | Established | 128K | $0.28 / $0.42 | 35 | 3.75s | 54.43Supported | |
| DeepSeek LLM 2.0 | Tracked | 128K | $0.00 / $0.00 | N/A | N/A | 54.03Estimated | |
| DeepSeek V3.1 | Established | 128K | $0.00 / $0.00 | N/A | N/A | 52.68Supported | |
| DeepSeek-R1 | Established | 128K | $0.55 / $2.19 | N/A | N/A | 50.74Supported | |
| DeepSeek Coder 2.0 | Tracked | 128K | N/A | N/A | N/A | 49.77Estimated | |
| DeepSeekMath V2 | Tracked | 128K | $0.00 / $0.00 | N/A | N/A | 49.38Estimated | |
| DeepSeek V3 | Established | 128K | $0.27 / $1.10 | N/A | N/A | 44.11Supported | |
| DeepSeek R1 Distill Qwen 32B | Established | 128K | $0.00 / $0.00 | 60 | 0.84s | 41.9Estimated | |
| DeepSeek V4 Pro (Max) | Current | 1M | $0.43 / $0.87 | N/A | N/A | — |
Radar
Full release historyDeepSeek release history
DeepSeek V4 Pro 0813 released
DeepSeek · Model release
Source confirmed
DeepSeek V4 Flash 0731 released
DeepSeek · Model release
Source confirmed
DeepSeek V4 released
DeepSeek · Model release
Source confirmed
DeepSeek V3.2 released
DeepSeek · Model release
Source confirmed
DeepSeek V3.1 released
DeepSeek · Model release
Source confirmed