MDL
DeepSeek V3 (Dec '24)
DeepSeek
Verified against Artificial Analysis · Jul 13, 2026
Capability
50
percentile index
Price /M
$0.4 / $0.89
in / out
Context
—
Avg score
37.6
11 benchmarks
AA Intelligence
14.2
index
Weights
Open
Released
Dec 26, 2024
About
DeepSeek V3 (Dec '24) is a open-weights model in the DeepSeek family from DeepSeek. Benchmarked on 11 evals, averaging 37.6.
Benchmark scores · 11
avg 37.6 — 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, 2026DeepSeek-V3 Technical Report223 cites · Dec 27, 2024OptSkills: Learning Generalizable Optimization Skills from Problem Archetypes via Cluster-Based DistillationMay 28, 2026FormInv: A Measurement Protocol for Semantic Invariance in Mathematical Reasoning BenchmarksMay 27, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Price in / out$0.4 / $0.89 per 1M
Blended price$0.523 / 1M
Coding index23.0
Math index26.0
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$0.89/M