MDL
DeepSeek V3.2
DeepSeek
Verified against Artificial Analysis · Jul 13, 2026
Capability
87
percentile index
Price /M
$0.28 / $0.42
in / out
Context
131.1K
tokens
Avg score
66.1
10 benchmarks
AA Intelligence
32.0
index
Weights
Open
Released
Dec 1, 2025
About
DeepSeek V3.2 is a open-weights model in the DeepSeek family from DeepSeek. Benchmarked on 10 evals, averaging 66.1.
Benchmark scores · 10
avg 66.1 — higher is betterThe research behind this model
all papers mentioning it →VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models▲ 104 on HF · Jun 15, 2026EEVEE: Towards Test-time Prompt Learning in the Real World for Self-Improving Agents▲ 18 on HF · Jun 9, 2026OptSkills: Learning Generalizable Optimization Skills from Problem Archetypes via Cluster-Based DistillationMay 28, 2026How Far Can Disaggregation Go? A Design-Space Exploration of Attention-FFN Disaggregation for Efficient MoE LLM ServingMay 27, 2026TCP-MCP: Landscape-Guided Co-Evolution of Prompts and Communication Topologies for Multi-Agent SystemsMay 27, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Price in / out$0.28 / $0.42 per 1M
Blended price$0.315 / 1M
Coding index44.2
Math index92.0
Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →
History
full ledger →Capability over time
Append-only ledger — every observed change to this model's numbers.
API providers
Artificial Analysis$0.42/M