DeepSeek
DeepSeek V3.2
Open SourceDeepSeek's open-weights frontier model and one of the most cost-effective APIs available. V3.2 punches far above its price — at $0.28/$1.10 per 1M tokens it costs roughly 20× less than Claude Sonnet while delivering an AA Intelligence Index of 32. Strong on coding and reasoning tasks, but hosted in China with the privacy implications that brings.
Context window
128K tokens
API (blended)
$0.49/1M
Consumer access
Free
Multimodal
Text only
Score Breakdown
40.3/100 → 4.0/10Intelligence, Reliability, Speed, and Context are field-relative — scores shift as models are added. Accessibility and Trust are absolute checklists. Full methodology →
Strengths
- +Exceptional price-to-performance: AA Intelligence Index 32 at ~$0.49/1M blended — roughly 12× cheaper than GPT-5.2
- +Open weights (MIT license) — self-hostable, auditable, no API dependency
- +Strong coding: LiveCodeBench 59.3% and τ²-bench 78.9% (AA-measured)
- +GPQA Diamond 75.1% (AA-measured) — competitive science reasoning for the price
- +Free consumer chat at chat.deepseek.com — no subscription required
- +Active open-source community and HuggingFace model support
Weaknesses
- -Chinese jurisdiction — data governed by PRC law; not suitable for sensitive enterprise data
- -128K context window is the smallest of any major frontier model
- -Speed: 43 t/s (AA-measured) — slower than most API-hosted competitors
- -No multimodal input (text only)
- -No MCP support
Best for
Not ideal for
Pricing details
Subscription plans
API pricing
Prices verified February 2026. LLM pricing changes frequently — verify at the provider's site before budgeting.
Last updated: February 27, 2026