losclouds

AI Models / Compare

DeepSeek V3

DeepSeek's price-performance family for coding, analytical chat, and structured tool workflows.

Creator
DeepSeek
Lifecycle
Active
Context
128.0K
Max output
8.2K
Released
Dec 26, 2025
Status
up
Input
$0.27 / 1M tokens
Output
$1.10 / 1M tokens
Cached read
— / 1M tokens
Cached write
— / 1M tokens
Batch discount
50%
Source
DeepSeek pricing
Verified
Apr 1, 2026 (Medium)

Capabilities

Modalities
text→text
Capabilities
reasoningbatchSupportfunctionCallingstructuredOutputs
Strengths
Low blended token cost, Strong coding value
Tradeoffs
Smaller host ecosystem than hyperscaler-backed families
Official Links
Intro
https://api-docs.deepseek.com
Docs
https://api-docs.deepseek.com
Pricing
https://api-docs.deepseek.com/quick_start/pricing
Release note
https://api-docs.deepseek.com/news/news

Benchmark Coverage

BenchmarkVersionScoreDateSourceNotes
HumanEval2026-Q191.2 pass@1Feb 18, 2026DeepSeek newsUse as raw evidence only.

Release History

ReleaseAliasLifecycleRelease DateDeprecationShutdownSummary
DeepSeek V3deepseek-v3ActiveDec 26, 2025——Current DeepSeek frontier family for production API workloads.

Host Coverage

HostTypeContextPricing NoteDifferences
DeepSeek APIfirst-party128.0KReference pricing and official model aliases.Direct vendor access
OpenRouteraggregator128.0KThird-party routing with model discovery and multi-provider failover.Aggregator routing
Migration Guidance

Strong price-performance destination for teams moving off more expensive general-purpose text families.

Breaking changes: Tooling and guardrail behavior differ from proprietary-provider ecosystems.

Change Events
DateTypeTitleDescriptionSource
Dec 26, 2025family_addedFamily launchedDeepSeek V3 launched as the vendor's current production API family.DeepSeek news

Other models from DeepSeek

DeepSeek Chat, DeepSeek Reasoner