deepseek
DeepSeek: R1 0528
deepseek/deepseek-r1-0528
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
deepseek/deepseek-r1-0528 is the May 2025 update to DeepSeek's open-weight reasoning model, released under an MIT license that permits commercial use and distillation. The OpenRouter catalog describes a large mixture-of-experts system with 671B total and 37B active parameters, text input and output, a roughly 164K-token context, and reasoning active on every request. DeepSeek reports major gains over the prior R1 in math, code, and logic benchmarks, plus reduced hallucination and better function calling; these are vendor-reported figures worth validating on your own workload.[1][2]
Strengths
- Vendor-reported evaluations show large improvements over the previous R1 across mathematics, competitive programming, and software-engineering benchmarks.[2]
- Open weights under an MIT license, with commercial use and distillation explicitly permitted by the model card.[2]
- Long context (about 164K tokens per the OpenRouter catalog) alongside tool calling, structured output support, and exposed reasoning tokens.[1][2]
Limitations
- Reasoning is mandatory per the OpenRouter catalog, and DeepSeek notes the model now thinks far longer per question, which can raise latency and token usage.[1][2]
- Most benchmark evidence is self-reported by DeepSeek or catalog-hosted; independent verification for this exact variant is limited, so validate on your workload.[1][2]
- OpenRouter's design-arena results show middling, near-even outcomes in visual and design categories such as SVG and game development.[1]
Where to use it
- Complex math, logic, and multi-step coding problems where visible reasoning traces help — suggested by vendor benchmarks, not independently confirmed here.[2]
- Self-hosted or distilled deployments: MIT licensing and a published 8B distillation make it a candidate for research and fine-tuning pipelines.[2]
- Agentic and tool-use workflows that need function calling and structured outputs, per the catalog's supported parameters.[1][2]
Sources
- [1] OpenRouter catalog · DeepSeek: R1 0528Sources checked: Nov 13, 2026
- [2] Model documentation · DeepSeek: R1 0528Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
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. It's 671B parameters in size, with 37B active...
Specifications
- Model ID
- deepseek/deepseek-r1-0528
- Creator
- deepseek
- Added to OpenRouter
- 28.05.2025
- Context length
- 163,840
- Modality
- text->text
- Input modalities
- text
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.500 / out $2.15
- Knowledge cutoff
- 31.03.2025
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | reasoning: mandatory · parameters: include_reasoning, reasoning |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: mandatory · parameters: include_reasoning, reasoning |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Design Arena | models · 3dwin rate 53.3% · rank 94 | 1,094Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · codecategorieswin rate 52.6% · rank 99 | 1,129Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · datavizwin rate 60.9% · rank 65 | 1,180Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · gamedevwin rate 49.3% · rank 102 | 1,088Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · svgwin rate 48.7% · rank 84 | 1,030Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 54.9% · rank 100 | 1,095Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · websitewin rate 52.7% · rank 97 | 1,142Elo | CC — below the B band | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| DeepInfrafp4 | 163,840 | $0.500 | $2.15 | — | — | 100% |
| SiliconFlowfp8 | 163,840 | $0.500 | $2.18 | — | — | 100% |
| StreamLake | 128,000 | $0.571 | $2.29 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.