DeepSeek R1
May 28th update to the [original DeepSeek R1](/deepseek/deepseek-r1) Performance on par with [OpenAI o1](/openai/o1), but open-sourced and with fully open reasoning tokens.
deepseek-r1Released 28 May 2025 (May 2025)Pricing
List prices at DeepSeek, in US dollars per million tokens.
| Tier | Input / 1M | Output / 1M | |
|---|---|---|---|
| Standard | $0.55 | $2.19 | |
| Cached input 75% off input | $0.14 | 75% off input |
What it costs a month
At DeepSeek list prices, 30 days.
- Chat assistant5,000 requests a day, 2K in / 0.5K out, cached$292
- RAG over documents2,000 requests a day, 8K in / 0.6K out, cached$244
- Coding agent1,000 requests a day, 40K in / 2K out, cached$447
- Overnight batch20,000 requests a day, 4K in / 0.4K out$1,846
Where to run it
18 hosts serve DeepSeek R1. The comparison is on blended price, 3 parts input to 1 part output.
| Host | In / out per 1M | Model id | |
|---|---|---|---|
DeepSeekLab API deepseek-r1 | $0.55 / $2.19 | deepseek-r1 | |
| $0.20 / $0.60 −69% vs headline | deepseek-r1-0528 | ||
HyperbolicInference DeepSeek-R1 | $0.40 / $0.40 −58% vs headline | DeepSeek-R1 | |
GMI CloudInference DeepSeek-R1 | $0.50 / $2.18 −4% vs headline | DeepSeek-R1 | |
Vercel AI GatewayGateway deepseek-r1 | $0.55 / $2.19 | deepseek-r1 | |
StreamLakeCloud deepseek-r1 | $0.571 / $2.29 +4% vs headline | deepseek-r1 | |
DeepInfraInference DeepSeek-R1 | $0.70 / $2.40 +17% vs headline | DeepSeek-R1 | |
Nebius AI StudioInference DeepSeek-R1 | $0.80 / $2.40 +25% vs headline | DeepSeek-R1 | |
Kyma APICloud deepseek-r1 | $0.743 / $2.96 +35% vs headline | deepseek-r1 | |
Azure AI FoundryCloud deepseek-r1 | $1.35 / $5.40 +146% vs headline | deepseek-r1 | |
SnowflakeCloud deepseek-r1 | $1.35 / $5.40 +146% vs headline | deepseek-r1 | |
| $1.35 / $5.40 +146% vs headline | deepseek-r1-0528-maas | ||
| $3.00 / $7.00 +317% vs headline | DeepSeek-R1-0528 | ||
| $3.00 / $7.00 +317% vs headline | DeepSeek-R1-0528 | ||
Novita AIInference deepseek-r1 | $4.00 / $4.00 +317% vs headline | deepseek-r1 | |
Fireworks AIInference deepseek-r1 | $3.00 / $8.00 +343% vs headline | deepseek-r1 | |
ReplicateInference deepseek-r1 | $3.75 / $10.00 +453% vs headline | deepseek-r1 | |
SambaNovaInference DeepSeek-R1 | $5.00 / $7.00 +473% vs headline | DeepSeek-R1 |
Show 6 more listings: regions, snapshots and variantsHide regional listings
| Hyperbolic | |||||||
| 0528 DeepSeek-R1-0528 | $0.25 / $0.25 −74% vs headline | DeepSeek-R1-0528 | |||||
| GMI Cloud | |||||||
| 0528 DeepSeek-R1-0528 | $0.57 / $2.29 +4% vs headline | DeepSeek-R1-0528 | |||||
| DeepInfra | |||||||
| 0528 DeepSeek-R1-0528 | $0.50 / $2.15 −5% vs headline | DeepSeek-R1-0528 | |||||
| Nebius AI Studio | |||||||
| 0528 DeepSeek-R1-0528 | $0.80 / $2.40 +25% vs headline | DeepSeek-R1-0528 | |||||
| Novita AI | |||||||
| 0528 deepseek-r1-0528 | $0.70 / $2.50 +20% vs headline | deepseek-r1-0528 | |||||
| Fireworks AI | |||||||
| 0528 deepseek-r1-0528 | $3.00 / $8.00 +343% vs headline | deepseek-r1-0528 | |||||
Capabilities
Core
- Streaming
- Function calling
- JSON mode
- Structured outputs
Multimodal
- Vision
- Image input
- Audio
- Image generation
- Video
Advanced
- Reasoning
- Tool calling
- MCP
- Prompt caching
- Batch API
- Embeddings
- Takes in
- Text
- Produces
- Text
- Features
- Assistant prefillPrompt caching
- Open weights
- License: DeepSeek License (MIT code)Runtimes: SGLang, TGI, vLLM
Cheaper alternatives
Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.
Same model, cheaper host
- Hyperbolic74% lessDeepSeek-R1-0528$0.25 / $0.25
- Lambda AI69% lessdeepseek-r1-0528$0.20 / $0.60
- Hyperbolic58% lessDeepSeek-R1$0.40 / $0.40
- DeepInfra5% lessDeepSeek-R1-0528$0.50 / $2.15
- GMI Cloud4% lessDeepSeek-R1$0.50 / $2.18
Smaller DeepSeek models
No cheaper DeepSeek model keeps its key capabilities.
Other frontier labs
- 11% less
- 52% less
- 79% less
- 79% less
Open-weight picks
- 9% less
- 10% less
- 17% less
- 29% less
Call it
Using DeepSeek's API.
from litellm import completion
response = completion(
model="deepseek/deepseek-r1",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)Quick answers
How much does DeepSeek R1 cost?
On DeepSeek, DeepSeek R1 costs $0.55 per million input tokens and $2.19 per million output tokens. Cached input costs $0.14 per million tokens. The cheapest host is Hyperbolic (0528), at $0.25 per million input tokens and $0.25 per million output tokens.
Which APIs offer DeepSeek R1?
DeepSeek R1 runs on 18 APIs: DeepSeek, Lambda AI, Hyperbolic, GMI Cloud, Vercel AI Gateway, StreamLake, DeepInfra, Nebius AI Studio, Kyma API and 9 more.
What is DeepSeek R1's context window?
66K tokens, with up to 8K tokens in one reply.
Is DeepSeek R1 being retired?
No retirement date is scheduled for it on any API we track.
What are cheaper alternatives to DeepSeek R1?
Models that cost less and keep most of what it can do: Gemini 3.5 Flash Lite on Google AI Studio (11% less), GPT-5.4 Nano on OpenAI (52% less) and Motif 3 on ai& (9% less). DeepSeek R1 itself costs 74% less on Hyperbolic (0528). Savings are on a mix of 3 input tokens to 1 output token.
When was DeepSeek R1 released?
DeepSeek released it on 28 May 2025.
What is DeepSeek R1's API model id?
Use deepseek-r1 with DeepSeek. Other hosts use their own ids: deepseek-r1-0528 on Lambda AI, DeepSeek-R1 on Hyperbolic, DeepSeek-R1 on GMI Cloud, DeepSeek-R1 on DeepInfra.
Sources
- GMI Cloud pricing
- Nebius AI Studio pricing
- SambaNova pricing
- Together AI pricing
- Azure AI Foundry pricing
- Fireworks AI pricing
- Novita AI pricing
- Replicate pricing
- StreamLake pricing
- Vertex AI pricing
- LiteLLM model map
Raw catalog entry for deepseek-r1 on DeepSeek
{
"cache_read_input_token_cost": 1.4e-7,
"input_cost_per_token": 5.5e-7,
"input_cost_per_token_cache_hit": 1.4e-7,
"litellm_provider": "deepseek",
"max_input_tokens": 65536,
"max_output_tokens": 8192,
"mode": "chat",
"output_cost_per_token": 0.00000219,
"supports_assistant_prefill": true,
"supports_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_tool_choice": true
}