LLM AtlasLLM AtlasSearch models

Provider analysis

DeepSeek

Lab publishing DeepSeek reasoning, coding, OCR, multimodal, and open model releases.

Last verified: 2026-03-29Confidence: HighPrimary sources: 3

This provider page blends full-profile entries with broader verified listings. Use it to separate deeply evaluated flagship models from source-backed records that are tracked primarily for market visibility, access data, and freshness coverage.

Headquarters
Hangzhou, China
Founded
2023
Models tracked
14
Full-profile models
14
Catalog last verified
2026-03-29
Latest model verification
2026-03-29
Newest release tracked
2025-03-01
Confidence
High
Access mix
open-weight, self-hosted, hosted
API models
0

Tracked models available through provider-managed APIs.

Open-weight models
14

Models with downloadable weights or self-hosted distribution paths.

Primary source links
50

Total source references attached across this provider catalog.

Provider sources

Official links used to verify the provider profile and platform coverage.

Last verified: 2026-03-29

DeepSeek

DeepSeek-V3.2

DeepSeek

DeepSeek's MoE flagship models (671B total, 37B activated params) with 128K context, strong on math, code, and reasoning benchmarks.

Score 714 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-V3.2-Exp

DeepSeek

DeepSeek's MoE flagship models (671B total, 37B activated params) with 128K context, strong on math, code, and reasoning benchmarks.

Score 714 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-V3.1

DeepSeek

DeepSeek's MoE flagship models (671B total, 37B activated params) with 128K context, strong on math, code, and reasoning benchmarks.

Score 714 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-V3.1-Base

DeepSeek

DeepSeek's MoE flagship models (671B total, 37B activated params) with 128K context, strong on math, code, and reasoning benchmarks.

Score 714 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-V3

DeepSeek

DeepSeek's MoE flagship models (671B total, 37B activated params) with 128K context, strong on math, code, and reasoning benchmarks.

Score 714 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-V2.5

DeepSeek

DeepSeek's MoE flagship models (671B total, 37B activated params) with 128K context, strong on math, code, and reasoning benchmarks.

Score 714 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-R1

DeepSeek R1

DeepSeek's reasoning model (671B total, 37B activated) trained with large-scale RL, achieving o1-level performance on math, code, and reasoning tasks.

Score 714 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-R1-Distill-Llama-70B

DeepSeek distill

DeepSeek-R1 distilled into Llama 3.3 70B, achieving strong reasoning with dense model efficiency.

Score 713 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-Coder-V2

DeepSeek Coder/Math

DeepSeek's specialist models for code generation and mathematical reasoning.

Score 713 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-Math-V2

DeepSeek Coder/Math

DeepSeek's specialist models for code generation and mathematical reasoning.

Score 713 sources
textreasoningcodeopen-sourceopen-weightself-hostedhosted
Context
128,000
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

Janus-Pro-7B

DeepSeek multimodal

DeepSeek's multimodal understanding and generation model (7B) for image-to-text and text-to-image tasks.

Score 704 sources
textvisionopen-sourceopen-weightself-hostedhosted
Context
16,384
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-VL2-Small

DeepSeek multimodal

DeepSeek's lightweight vision-language model for efficient multimodal understanding.

Score 703 sources
textvisionopen-sourceopen-weightself-hostedhosted
Context
16,384
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-OCR

DeepSeek multimodal

DeepSeek's specialized OCR models for document parsing, table extraction, and image-to-text conversion.

Score 703 sources
textvisionopen-sourceopen-weightself-hostedhosted
Context
16,384
Input
Not applicable
Output
Not applicable
Coverage
Full profile

DeepSeek

DeepSeek-OCR-2

DeepSeek multimodal

DeepSeek's specialized OCR models for document parsing, table extraction, and image-to-text conversion.

Score 703 sources
textvisionopen-sourceopen-weightself-hostedhosted
Context
16,384
Input
Not applicable
Output
Not applicable
Coverage
Full profile