DeepSeek V3.2 by DeepSeek demonstrates strong general intelligence, excellent coding ability, competitive pricing. View detailed benchmark data including scores across coding, math, reasoning, speed, and cost metrics.
Specifications last checked against the provider on . Benchmark scores carry their own source, linked beside each number.
General Benchmarks
Massive Multitask Language Understanding professional benchmark
HuggingFace LeaderboardHuman preference ELO from blind head-to-head votes
LMSYS / HuggingFaceCoding Benchmarks
Human preference ELO for building a working web page from a brief
Design Arena (via OpenRouter)Artificial Analysis composite coding score
Artificial Analysis (via OpenRouter)Human preference ELO for building things — websites, UI, games, charts, SVG
Design Arena (via OpenRouter)Math Benchmarks
Reasoning Benchmarks
Multi-turn service agent making tool calls under strict policy constraints
OpenRouter (measured)Speed Benchmarks
Cost Benchmarks
Cost per 1M cached input tokens — the price that actually applies to a long agent conversation
OpenRouter APIContext Benchmarks
Available from 34 providers
The same model costs different amounts depending on who serves it. You can bring your own key for 7 of these — connect it here.
| Provider | Input / 1M | Output / 1M | Cached in | Context | Your key |
|---|---|---|---|---|---|
| Alibaba Token Plan | $0 | $0 | — | 131K | — |
| Alibaba Token Plan (China) | $0 | $0 | — | 131K | — |
| iFlow | $0 | $0 | — | 128K | — |
| CrofAI | $0.18 | $0.35 | $0.04 | 164K | — |
| TokenGo | $0.2174 | $0.326 | $0.06 | 128K | — |
| ZenMux | $0.22 | $0.33 | — | 163K | — |
| Deep Infra | $0.26 | $0.38 | $0.13 | 164K | Supported |
| DevPass (LLM Gateway) | $0.26 | $0.38 | $0.13 | 164K | — |
| LLM Gateway | $0.26 | $0.38 | $0.13 | 160K | — |
| Meganova | $0.26 | $0.38 | — | 164K | — |
| Kilo Gateway | $0.269 | $0.4 | $0.1345 | 164K | — |
| Abacus | $0.27 | $0.4 | — | 128K | — |
| Helicone | $0.27 | $0.41 | — | 164K | — |
| NovitaAI | $0.27 | $0.41 | — | 164K | — |
| OpenRouter | $0.27 | $0.41 | — | 164K | Supported |
| Poe | $0.27 | $0.4 | $0.13 | 128K | — |
| Hugging Face | $0.28 | $0.4 | — | 164K | Supported |
| Merge Gateway | $0.28 | $0.45 | $0.14 | 164K | — |
| NanoGPT | $0.28 | $0.42 | $0.14 | 163K | — |
| Vivgrid | $0.28 | $0.42 | — | 128K | — |
| Alibaba (China) | $0.287 | $0.431 | — | 131K | — |
| 302.AI | $0.29 | $0.43 | — | 128K | — |
| Ofox | $0.29 | $0.43 | $0.06 | 128K | — |
| Cortecs | $0.296 | $0.495 | $0.075 | 164K | — |
| TensorX | $0.3 | $0.5 | $0.075 | 164K | — |
| Venice AI | $0.33 | $0.48 | $0.16 | 160K | — |
| Friendli | $0.5 | $1.5 | $0.25 | 164K | — |
| Vultr | $0.55 | $1.65 | — | 131K | — |
| Vertex | $0.56 | $1.68 | $0.056 | 164K | Supported |
| EmpirioLabs AI | $0.57 | $1.71 | $0.57 | 128K | — |
| Azure | $0.58 | $1.68 | — | 128K | Supported |
| Azure Cognitive Services | $0.58 | $1.68 | — | 128K | — |
| Amazon Bedrock | $0.62 | $1.85 | — | 164K | Supported |
| Vercel AI Gateway | $0.62 | $1.85 | — | 128K | Supported |
DeepSeek V3.2 — Benchmark Scores Overview
Scores normalized to percentage scale for visual comparison. ELO scores mapped to 0-100 range (1100-1500).
Compare DeepSeek V3.2 With
DeepSeek V3.2 — Frequently Asked Questions
How intelligent is DeepSeek V3.2?
DeepSeek V3.2 scores 1370 on the Chatbot Arena ELO rating, making it a high-performing AI model. This score is based on blind head-to-head human preference voting.
How much does DeepSeek V3.2 cost?
DeepSeek V3.2 costs $0.27 per 1M input tokens and $0.40 per 1M output tokens. This makes it one of the more affordable models.
How fast is DeepSeek V3.2?
DeepSeek V3.2 generates output at 70 tokens per second, which is slower, prioritizing quality over speed compared to other models. The time to first token is 400 ms.
How good is DeepSeek V3.2 at coding?
DeepSeek V3.2 achieves 73.0% on SWE-bench Verified, demonstrating excellent real-world software engineering capability. This benchmark tests the model's ability to resolve actual GitHub issues.
How good is DeepSeek V3.2 at math and reasoning?
DeepSeek V3.2 scores 85.0% on the MATH benchmark (competition-level mathematics). It also achieves 78.9% on GPQA Diamond, a graduate-level science reasoning benchmark.
What is the context window of DeepSeek V3.2?
DeepSeek V3.2 has a context window of 164K tokens. This determines how much text, conversation history, and code the model can process in a single request.
Who created DeepSeek V3.2?
DeepSeek V3.2 was created by DeepSeek. It is classified as a open source model in our catalogue.
Is DeepSeek V3.2 open source?
Yes, DeepSeek V3.2 is an open-source model. The model weights are publicly available for download and self-hosting.