---
title: DeepSeek
slug: deepseek
url: "https://toolweight.com/tools/deepseek"
homepage: "https://www.deepseek.com"
categories: llm-apis
last_verified: 2026-07-30
dataset_version: 0738be72
license: CC-BY-4.0
---

# DeepSeek

> Frontier-adjacent models at a small fraction of Western prices

- Reset the category's price floor and has kept cutting since
- V3 line released under MIT, so the same model runs on your hardware or a Western host
- The cheapest serious first-party API here
- Governance terms are the most permissive, the reason to route around it for sensitive workloads

## Identity

|  |  |
| --- | --- |
| Name | DeepSeek |
| Company | DeepSeek |
| One-liner | Frontier-adjacent models at a small fraction of Western prices |
| Site | https://www.deepseek.com |
| Docs | https://api-docs.deepseek.com |
| Founded | 2023 |
| Open source | Yes |
| Licence | MIT |
| Compared in | 1 |

## Where it is compared

### [LLM APIs](https://toolweight.com/compare/llm-apis)
Ranked **#18 of 18** on default weights.
| Field | Value | Confidence | Verified | Source | Note | Permalink |
| --- | --- | --- | --- | --- | --- | --- |
| Flagship model | DeepSeek-V4-Flash (deepseek-v4-flash) | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/quick_start/pricing | The `deepseek-chat` / `deepseek-reasoner` aliases were deprecated 2026-07-24 and now route to v4-flash (non-thinking and thinking modes); `deepseek-v4-pro` is the higher-capability tier. Aliases still change the model behind them without a version bump. | https://toolweight.com/compare/llm-apis#deepseek--flagship_model |
| Context window | 1,000,000 tokens | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/news/news260424/ | 1M context is now the default across all official DeepSeek services. | https://toolweight.com/compare/llm-apis#deepseek--context_window |
| Max output | 384,000 tokens | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/quick_start/pricing | - | https://toolweight.com/compare/llm-apis#deepseek--max_output |
| Image input | ○ | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/news/news260424/ | The V4 models are text-only; vision needs the separate DeepSeek-VL models. | https://toolweight.com/compare/llm-apis#deepseek--vision |
| Audio in/out | ○ | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/news/news260424/ | V4 is text in, text out. | https://toolweight.com/compare/llm-apis#deepseek--audio_io |
| Open weights | ● | Vendor-claimed | 2026-01-15 | https://github.com/deepseek-ai/DeepSeek-V3 | V3 line released under MIT, among the most permissive licences of any frontier-adjacent model. | https://toolweight.com/compare/llm-apis#deepseek--open_weights |
| OpenAI-compat API | ● | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/news/news260424/ | Supports the OpenAI ChatCompletions and Anthropic APIs. | https://toolweight.com/compare/llm-apis#deepseek--openai_compatible |
| TTFT p50 | - | Unknown | - | - | - | https://toolweight.com/compare/llm-apis#deepseek--ttft_p50 |
| Output tok/s | - | Unknown | - | - | - | https://toolweight.com/compare/llm-apis#deepseek--throughput_tps |
| $/M input | $0.14/M tok | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/quick_start/pricing | v4-flash cache-miss rate; a cache hit bills $0.0028/M. The higher-capability v4-pro is $0.435/M miss. | https://toolweight.com/compare/llm-apis#deepseek--price_in |
| $/M output | $0.28/M tok | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/quick_start/pricing | v4-flash output rate; v4-pro is $0.87/M. Still a small fraction of the frontier volume tiers — which ratio is the honest one depends on which model you would otherwise have used. | https://toolweight.com/compare/llm-apis#deepseek--price_out |
| $/M cache read | $0.0028/M tok | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/quick_start/pricing | Cache hits bill at a fiftieth of the v4-flash miss rate, applied automatically. | https://toolweight.com/compare/llm-apis#deepseek--cached_input |
| Batch discount | 0% | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/quick_start/pricing | No batch endpoint on the official pricing page; the list price is already below most competitors' batch rates. | https://toolweight.com/compare/llm-apis#deepseek--batch_discount |
| Tool use | ● | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/news/news260424/ | Function calling with dedicated agent-capability optimisations in V4; multi-step reliability has historically been reported as weaker than the closed frontier. | https://toolweight.com/compare/llm-apis#deepseek--tool_use |
| Schema output | ◐ | Inferred | 2026-01-15 | - | JSON mode rather than full schema-constrained decoding. | https://toolweight.com/compare/llm-apis#deepseek--structured_output |
| Effort control | ◐ | Inferred | 2026-01-15 | - | A separate reasoner model rather than a per-request effort level. | https://toolweight.com/compare/llm-apis#deepseek--reasoning_effort |
| Computer use | ○ | Inferred | 2026-01-15 | - | - | https://toolweight.com/compare/llm-apis#deepseek--computer_use |
| MCP support | ○ | Inferred | 2026-01-15 | - | - | https://toolweight.com/compare/llm-apis#deepseek--mcp_support |
| Cache TTL | Automatic disk cache, no fixed TTL | Vendor-claimed | 2026-07-30 | https://api-docs.deepseek.com/news/news0802/ | Context caching on disk; unused entries are cleared automatically, typically within a few hours to days, with no user configuration. | https://toolweight.com/compare/llm-apis#deepseek--prompt_cache_ttl |
| Continuity policy | The worst on this page for a closed endpoint, and the best if you self-host. The API offers only rolling aliases, deepseek-chat silently points at whatever the current model is, so behaviour can change with no version to pin and no deprecation notice. The mitigation is the MIT licence: download the checkpoint and the version is yours permanently. | Community-reported | 2026-01-15 | - | - | https://toolweight.com/compare/llm-apis#deepseek--continuity_policy |
| Notice period | - | Unknown | - | - | No published deprecation policy; aliases are updated in place. | https://toolweight.com/compare/llm-apis#deepseek--deprecation_notice |
| Pinnable versions | ○ | Community-reported | 2026-01-15 | - | Only rolling aliases are exposed on the first-party API. | https://toolweight.com/compare/llm-apis#deepseek--pinned_snapshots |
| Zero retention | ○ | Inferred | 2026-01-15 | - | - | https://toolweight.com/compare/llm-apis#deepseek--zero_retention |
| Trains on your data | ● | Community-reported | 2026-01-15 | - | Terms permit using inputs to improve services. Assume prompts are retained and read them before sending anything sensitive. | https://toolweight.com/compare/llm-apis#deepseek--trains_on_data |
| AU region | ○ | Inferred | 2026-01-15 | - | - | https://toolweight.com/compare/llm-apis#deepseek--au_region |
| API since | 2023 | Inferred | - | - | - | https://toolweight.com/compare/llm-apis#deepseek--launch_year |
| Positioning | Near-frontier quality at a fraction of the token price | Inferred | - | - | - | https://toolweight.com/compare/llm-apis#deepseek--positioning |

**Verdict.** The category's price floor with unmatched quality-per-dollar for bulk work; for regulated data, use the MIT weights via a Western host.

## Alternatives

- [Qwen](https://toolweight.com/tools/alibaba-qwen), Alibaba's model family, huge open-weight range, closed flagship
- [Amazon Bedrock](https://toolweight.com/tools/amazon-bedrock), Multi-vendor model access inside your existing AWS account
- [Anthropic](https://toolweight.com/tools/anthropic), Claude models, built around long agentic runs and tool use
- [Cerebras](https://toolweight.com/tools/cerebras), Wafer-scale inference, the fastest tokens per second available
- [Cohere](https://toolweight.com/tools/cohere), Enterprise-focused models built for RAG and private deployment
- [Fireworks AI](https://toolweight.com/tools/fireworks-ai), Fast open-weight inference with strong structured-output support
- [Google Gemini](https://toolweight.com/tools/google-gemini), Gemini via AI Studio for prototyping or Vertex AI for production
- [Groq](https://toolweight.com/tools/groq), Custom LPU silicon serving open-weight models at extreme speed
- [Meta Llama](https://toolweight.com/tools/meta-llama), Open-weight Llama models, hosted almost everywhere but Meta
- [Mistral AI](https://toolweight.com/tools/mistral-ai), European lab with an open-weight lineage and EU-resident hosting
- [Moonshot AI](https://toolweight.com/tools/moonshot-ai), Kimi models, open-weight agentic performance at low cost
- [OpenAI](https://toolweight.com/tools/openai), GPT models plus audio, images and embeddings on one bill

## Logo & brand colours

SVG files: https://toolweight.com/teams/deepseek/logo.svg and https://toolweight.com/teams/deepseek/mono.svg. JSON: https://toolweight.com/api/v1/brands/deepseek.

| Hex | oklch |
| --- | --- |
| #4d6bfe | oklch(59.2% 0.220 269.6) |

## Licence and attribution

Data from Toolweight (https://toolweight.com), licensed CC-BY-4.0.

- Licence: [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)
- Canonical HTML: https://toolweight.com/tools/deepseek
- Machine-readable: https://toolweight.com/tools/deepseek.md · https://toolweight.com/api/v1 · https://toolweight.com/mcp
- Toolweight takes no affiliate revenue and sells no placements.