Model

DeepSeek-V3.1 (Thinking)

OPEN

DeepSeek · DeepSeek-V3 family · released Aug 21, 2025

Thinking mode of the hybrid V3.1 checkpoint; served via deepseek-reasoner, faster than R1-0528 at similar quality

ReasoningCodingVisionFunction callingTool useAgentic

1903.9

Elo · rank #66

Parameters

671B

Active params

37B (MoE)

Context

128K tokens

Architecture

Mixture-of-Experts (671B total / 37B active) — DeepSeekMoE + Multi-head Latent Attention, hybrid think/non-think chat template

License

MIT

Languages

—

API price (in/out)

$0.56 / $1.68

Modalities

text

Benchmark results

Bar shows position within the tracked field; marker = field best

Aider PolyglotCoding76.3%#8

best: Claude Opus 4.5 · 89.4%

AIMEMath93.1%#29

best: GPT-5.2 · 100.0%

BrowseCompAgents30.0%#38

best: Kimi K3 · 91.2%

CodeforcesCoding2091#16

best: DeepSeek-V4-Pro (Think Max) · 3206

GPQA DiamondReasoning80.1%#78

best: GPT-5.6 · 94.6%

Humanity's Last ExamReasoning15.9%#72

best: Claude Sonnet 5 · 57.4%

LiveCodeBenchCoding74.8%#57

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MMLU-ProKnowledge84.8%#28

best: Claude Fable 5 · 91.5%

MMLU-ReduxKnowledge93.7%#10

best: Qwen3.7-Max · 95.0%

Run it locally

VRAM @ Q4

405 GB

VRAM @ FP16

1340 GB

Fits on (Q4)

Multi-node cluster required

Exceeds a single RTX 4090's 24GB VRAM even at Q4 (~405GB); requires multi-GPU or heavy CPU offload

Quantizations

GGUF · FP8

DeepSeek-V3 family

Elo progression across releases

DeepSeek-V3 671BDec 20241269.5 DeepSeek-V3-0324Mar 20251365.5 DeepSeek-V3.1 (Non-Thinking)Aug 20251538.3 DeepSeek-V3.1 (Thinking)Aug 20251903.9 DeepSeek-V3.2-ExpSep 20251875.1

Resources

HFdeepseek-ai/DeepSeek-V3.1 GHdeepseek-ai/DeepSeek-V3 DOCSapi-docs.deepseek.com APIAvailable via DeepSeek API

API price $0.56/$1.68 · each benchmark row carries its own source badge (see methodology)