MODEL PROOF
DeepSeek V3.2 Exp
BETTER VALUE OPTION
A stored alternative has an equal-or-higher measured score and an equal-or-lower price.
DeepSeek V4.1 Flash · Δ score 40.4 points · 40% lower measured price · $0.182 / 1M tokens (Price under confirmation since Oct 12, 2026)
GLM 5.3 Flash is a further stored option with these losses: 147K → 131K max output
- Independent LMArena score
- 1,422.1 ± 6.34 · 12,005 votes
- Context window
- 163.84K
- Maximum output
- 147.46K
- Input modalities
- text
- Output modalities
- text
- Published input price
- $0.270 / 1M tokens
- Published output price
- $0.410 / 1M tokens
- Pricing kind
- fixed
Inspect complete billing conditions and endpoint terms below.
Price from SiliconFlow: the cheapest offer that is serving, at standard delivery, with at least 90% of the reference context and at a declared precision that is not 4-bit (fp8), at its standard rate. The maker’s own precision is not known.
Every price dimension · USD per million tokens
| Input | $0.270/M |
|---|---|
| Output | $0.410/M |
| Cached input | Unknown/M |
| Cache write | Unknown/M |
| Cache write 1h | Unknown/M |
| Reasoning | Unknown/M |
Who sells it
The same weights, different shops. Cheapest is not like-for-like when serving precision differs.
1 provider offers · rates, limits and conditions
Rates are USD per million tokens. Endpoint terms can differ even when the quoted price and precision match. A listed parameter is a provider declaration, not a task-success test.
| Seller | Input | Output | Precision | Uptime (1d) |
|---|---|---|---|---|
| SiliconFlowsiliconflow/fp8 Endpoint terms
| $0.270 | $0.410 | fp8 | 99.86% |
Independent scores
| LMArena Elo | 1422.1default |
|---|---|
| LMArena row | deepseek-v3.2-exp |
LMArena Elo under CC BY 4.0. Absence of another board is not a score of zero.
Best rankings by task
- 3D #67 1149
- UI components #68 1166
- Code categories #70 1172
- Game development #71 1148
- Websites #72 1180
- SVG #78 1031
- Data visualisation #79 1152
Per-task Elo and rank from Design Arena, via OpenRouter’s model feed. The rank is the feed’s, and it counts models OpenRouter does not list.
Capability
| Context window | 163K |
|---|---|
| Max output | 147K |
| Input modes | text |
| Tool use | yes |
| Reasoning | optional |
| Knowledge cutoff | 2025-07-31 |
| Open weights | Unknown |
Provenance
| Price source | openrouter.ai |
|---|---|
| Fetched | 2026-10-16 |
| Quality data | verified |
DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...
Questions this page answers
What does DeepSeek V3.2 Exp cost?
$0.270/M in, $0.410/M out in this catalogue, as of the fetch date on this page. That is the API row, not a subscription.
Does DeepSeek V3.2 Exp have an independent quality score?
DeepSeek V3.2 Exp has an LMArena score in this catalogue.
What beats DeepSeek V3.2 Exp?
DeepSeek V4.1 Flash has an equal-or-higher measured score and an equal-or-lower price.
Does this page use Artificial Analysis scores for DeepSeek V3.2 Exp?
No. Artificial Analysis figures are not published here. Quality on this page is LMArena Elo where a score exists; otherwise the row is unrated.
MODEL MONUMENT
Save or share this proof
Badge
The verdict as one SVG, for a README or a docs page. It states this model's dominance status at the balanced workload on the LMArena lens, with the date it was computed, and it is rebuilt with the catalogue — so it changes when the verdict changes, including to one you would rather it did not.
[](https://undominated.ai/models/deepseek__deepseek-v3.2-exp/)<a href="https://undominated.ai/models/deepseek__deepseek-v3.2-exp/"><img src="https://undominated.ai/badge/deepseek__deepseek-v3.2-exp.svg" alt="DeepSeek V3.2 Exp — Undominated.ai dominance verdict" height="36"></a>Direct file: https://undominated.ai/badge/deepseek__deepseek-v3.2-exp.svg — a static SVG, written by scripts/build-badges.mjs on every build, so the copy you embed is never
older than the last deploy.
