DeepSeek R1
Active76.9StrongOpen-weight reasoning model matching o1 performance, fully open-source.
Intelligence Index
76.9/ 100
Strongweighted across 5 benchmarks
- General knowledge
- 90.8
- Math
- 88.5
- Reasoning
- 71.5
- Coding
- 56.9
Computed as the mean of per-category averages across MMLU, GPQA, SWE-bench, HumanEval, MATH, GSM8K, AIME, Aider Polyglot and more. See each benchmark for methodology.
Overview
DeepSeek R1 is a reasoning model trained with reinforcement learning to produce chain-of-thought before answering. It matches or exceeds OpenAI o1 on several math and coding benchmarks and is available as a fully open-weight model.
History
DeepSeek R1 became available via the DeepSeek API on 2025-01-20.
Training & availability
Training data has a knowledge cutoff of 2024-07-31 — information about events after that date is unlikely to appear in the model's responses. Weights are publicly available under the MIT license, making this an open-weight model suitable for on-prem deployment and fine-tuning.
Capabilities
-
Context window: 128K tokens.
-
Max output: 32K tokens.
-
Input modalities: text.
-
Intelligence Index: 76.9/100.
Strongest categories: General knowledge (91), Math (89), Reasoning (72).
Recommended for: math, open-source, reasoning.
Limitations
-
The knowledge cutoff is 20 months old — this model will not know about recent events, releases, or API changes.
-
Text-only — cannot process images, audio, or video inputs.
Pricing
- Input: $0.5500 per 1M tokens
- Output: $2.1900 per 1M tokens
Use the cost calculator above to estimate monthly spend for your workload.
Quick start
Minimal example using the OpenRouter API. Copy, paste, replace the key.
from openai import OpenAI
client = OpenAI(
base_url="https://openrouter.ai/api/v1",
api_key="sk-or-...",
)
resp = client.chat.completions.create(
model="deepseek/deepseek-r1",
messages=[{"role": "user", "content": "Explain quantum computing in one sentence."}],
)
print(resp.choices[0].message.content)Cost calculator
Estimate your monthly bill. Presets are typical workload sizes.
Providers & performance
2 providersMulti-provider inference routes for this model — sorted by throughput. Latency is time-to-first-token; throughput is output tokens per second. Data from OpenRouter, measured over the last 30 minutes.
| Provider | Throughput | Latency (TTFT) | Input $ / 1M | Output $ / 1M | Context | Quant | Supports |
|---|---|---|---|---|---|---|---|
| Azure | 67tok/s | 1.41s | $1.49 | $5.94 | 164K | — | — |
| Novita | 30tok/s | 1.63s | $0.7 | $2.5 | 64K | fp8 | tools |
Popularity
Signals from open-source communities — not a quality measure, but useful for gauging adoption among developers.
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| AIME 2024Math | 79.8% accuracy | Self-reported DeepSeek R1 paper |
| Aider PolyglotCoding | 56.9% pass@2 | Third-party Papers With Code |
| GPQA DiamondReasoning | 71.5% accuracy | Self-reported DeepSeek R1 paper |
Integrations & tooling support
- Tool calling
- Not supported
- Structured outputs
- Not supported
Price vs quality
Mid-tier performance and pricing — standard choice.
- Quality percentile
- 68.1%
- Effective price
- $1.78/1M
- Pricing breakdown
- $0.55/1M in
$2.19/1M out
Community ratings
Rate DeepSeek R1
Sign in to rate and review.
Comments
Sign in to leave a comment.