DeepSeek-V3.2-Exp: API Pricing, Context Window & Benchmarks
DeepSeek-V3.2-Exp is a language model from DeepSeek, released in September 2025.
DeepSeek-V3.2-Exp is an experimental iteration introducing DeepSeek Sparse Attention (DSA) to improve long-context training and inference efficiency while keeping output quality on par with V3.1. It explores fine-grained sparse attention
DeepSeek-V3.2-Exp benchmarks
Capability tiers
Standing within each category, adjusted for leaderboard depth.
Real tasks performance
High-confidence performance for DeepSeek-V3.2-Exp across real-world prompt categories. Only 95% intervals at most 4 points wide are shown.
Performance by conversation depth
How DeepSeek-V3.2-Exp holds up as conversations get longer.
Quality Tracker
DeepSeek-V3.2-Exp Performance Across Datasets
Scores sourced from the model's scorecard, paper, or official blog posts
llm-stats.com - Tue Sep 08 2026DeepSeek-V3.2-Exp pricing
Providers
DeepSeek-V3.2-Exp starts at $0.270 per million input tokens and $0.410 per million output tokens via Novita.
| Provider | Input $/M | Cached input $/M | Output $/M | Context in / out | TTFT p95 s | Output p5 c/s | Modalities in / out |
|---|---|---|---|---|---|---|---|
 Novita | $0.270 | — | $0.410 | 163.8K/65.5K | — | — | / |
Cached input is the discounted price for prompt tokens served from a provider cache. TTFT is time to first token. Output is characters per second; p5 is the sustained floor exceeded by 95% of observed requests.
DeepSeek-V3.2-Exp model size
DeepSeek-V3.2-Exp has 685 billion parameters. See how it compares to other models in the same parameter range.
DeepSeek-V3.2-Exp context window
Input and output token limits for DeepSeek-V3.2-Exp, plus how it ranks on long-context understanding.
DeepSeek-V3.2-Exp API
Available from the model provider
DeepSeek-V3.2-Exp has an official provider API. It is not currently routed through the LLM Stats gateway.
Read the official API documentationDeepSeek-V3.2-Exp latency
DeepSeek-V3.2-Exp time to first token, sustained output throughput, and failed-request rate from live API traffic over the trailing 7 days.
DeepSeek-V3.2-Exp examples
Recent arena outputs from DeepSeek-V3.2-Exp, picked from the highest-ranked matchups.
DeepSeek-V3.2-Exp license
DeepSeek-V3.2-Exp is released under the MIT license, which permits commercial use, has 685.0B parameters.
- License
- MIT
- Commercial use allowed
- Parameters
- 685.0B
MIT License - allows commercial use
DeepSeek-V3.2-Exp resources
Official sources for DeepSeek-V3.2-Exp: api documentation, official playground, paper or system card, source repository, model weights.
DeepSeek-V3.2-Exp vs other models
The most-compared alternatives to DeepSeek-V3.2-Exp are GPT-5 Medium, EXAONE 4.5 33B, Gemini 2.5 Pro Preview 06-05. Open any pair side-by-side for benchmarks, pricing, context, and latency.
Models like DeepSeek-V3.2-Exp
Models ranked just above and below DeepSeek-V3.2-Exp by LLM Stats score.
FAQ
Common questions about DeepSeek-V3.2-Exp.