AI model
deepseek/deepseek-v3.2
DeepSeek: DeepSeek V3.2 is a 163,840-token model served on Composite at $0.261 per million tokens in and $0.387 per million tokens out. Ranked #36 of 179 models by traffic on Composite over the last 30 days, taking 0.38% of tokens and 0.38% of requests (this model and its routing variants). 1 up and 0 down from 1 vote. A positive score is published once a model reaches 5 votes. Measured on Composite, 99.0% of its last 100 requests on Composite succeeded, the typical first token arrives in 8690 ms, output runs at about 134.9 tokens per second.
Try in Playground| Model ID | deepseek/deepseek-v3.2 |
|---|---|
| Provider | cv11 |
| Context window | 163,840 tokens |
| Accepts | text |
| Returns | text |
| Input price | $0.261 per million tokens |
| Output price | $0.387 per million tokens |
| Prompt caching | Supported — cached input is billed at the provider's cache rate, on credits and plans alike |
| Moderated | No |
| Cost on a monthly plan | About 2.0 plan credits per typical call (8k in, 1k out) |
| Routing endpoints | 10 (base plus 9 routing variants) |
Make a Composite API key and type deepseek/deepseek-v3.2 as the model name. JanitorAI takes the full URL https://composite.lucidity.sh/v1/chat/completions; SillyTavern, ChubAI, Agnai and RisuAI take https://composite.lucidity.sh/v1. It runs on credits, a plan, or your own provider key.
JanitorAI setup · SillyTavern setup · Best Janitor AI proxies compared
#36 of 179
0.38% of tokens and 0.38% of requests over 30 days, including routing variants.
99.0%
Across the last 100 requests Composite served for this model.
8690 ms
Median across the same sample.
134.9 tok/s
Median generation rate after the first token.
40.4%
Share of prompt tokens served from the provider's cache over 7 days.
The same model behind 9 alternative endpoints. Call the id directly to pin a route; the base id above picks for you.
| Route | Model ID | Provider | Price |
|---|---|---|---|
| Normal | deepseek/deepseek-v3.2-normal |
cv11 | $0.325 per million tokens input · $0.475 per million tokens output |
| Normal + cheapThink | deepseek/deepseek-v3.2-normal:cheapThink |
cv11 | $0.325 per million tokens input · $0.475 per million tokens output |
| Cheapest provider | deepseek/deepseek-v3.2:cheapest-provider |
GMICloud | $0.261 per million tokens input · $0.387 per million tokens output |
| Cheapest provider + cheapThink | deepseek/deepseek-v3.2:cheapest-provider:cheapThink |
GMICloud | $0.261 per million tokens input · $0.387 per million tokens output |
| cheapThink | deepseek/deepseek-v3.2:cheapThink |
cv11 | $0.261 per million tokens input · $0.387 per million tokens output |
| Fast | deepseek/deepseek-v3.2:fast |
GMICloud | $0.261 per million tokens input · $0.387 per million tokens output |
| Fast + cheapThink | deepseek/deepseek-v3.2:fast:cheapThink |
GMICloud | $0.261 per million tokens input · $0.387 per million tokens output |
| Quality | deepseek/deepseek-v3.2:quality |
GMICloud | $0.261 per million tokens input · $0.387 per million tokens output |
| Quality + cheapThink | deepseek/deepseek-v3.2:quality:cheapThink |
GMICloud | $0.261 per million tokens input · $0.387 per million tokens output |
$0.261 per million tokens input · $0.387 per million tokens output
Credits pay the provider’s price plus 25%; cache hits cost less.
On a monthly plan, a typical call (8k tokens in, 1k out) uses about 2.0 plan credits, at the provider's own price.
cv11
DeepSeek models are known for strong reasoning and coding ability at aggressive pricing, which makes them a common pick for roleplay accounts that send a high volume of messages.
With a 163,840-token context window, it can hold a long-running roleplay or a full lorebook in memory without losing earlier plot details, so it suits multi-session campaigns and character cards with heavy backstory.
It is priced low enough for high-volume daily roleplay without the per-message cost adding up quickly.
DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...
Supplied by cv11, not written by Composite.
DeepSeek: DeepSeek V4.1 Flash 1017 · DeepSeek: DeepSeek V4.1 Flash · DeepSeek: DeepSeek V4 Flash Vision Exp