DeepSeek V3.2 (chat) vs Llama 3.3 70B Instruct

Llama 3.3 70B Instruct is cheaper than DeepSeek V3.2 (chat) by about 51% on a blended (75% input / 25% output) basis. DeepSeek V3.2 (chat) charges $0.28 / $0.42 per 1M input/output tokens; Llama 3.3 70B Instruct charges $0.10 / $0.32.

FieldDeepSeek V3.2 (chat)Llama 3.3 70B Instruct
Input $/Mtok$0.28$0.10
Output $/Mtok$0.42$0.32
Cache read $/Mtok$0.0280–
Cache write $/Mtok––
Input $/Mchar0.09180.0301
Output $/Mchar0.140.0963
Context window128000128000
Tokenizerdeepseekllama3
Confidence?aggregatorapi
Sourcelitellmopenrouter

Confidence: official = vendor pricing page · api = OpenRouter per-host · aggregator = community list, may lag · unreviewed = automatic extraction awaiting review. When sources disagree the most authoritative wins. Details: /methodology#confidence

Price history

No price changes recorded yet for either model.

Same workload, monthly cost

Illustrative monthly cost at 10M tokens/month, split per workload preset. Not a substitute for the calculator, which uses your own volume, cache and batch assumptions.

WorkloadDeepSeek V3.2 (chat)Llama 3.3 70B Instruct
Coding agent$2.87$1.11
RAG chat$3.01$1.33
Summarisation (batch)$2.94$1.22
Classification$2.83$1.04
Chat assistant$3.36$1.88
Content generation$3.92$2.76

Batch pricing not published for DeepSeek V3.2 (chat); standard rate used for the batch share of a workload.

Batch pricing not published for Llama 3.3 70B Instruct; standard rate used for the batch share of a workload.

Cite this page

DeepSeek V3.2 (chat) vs Llama 3.3 70B Instruct: $0.28/$0.42 vs $0.10/$0.32 per 1M tokens as of 2026-11-19. Token Barometer by Dejan Strbad, https://tokenbarometer.com, data CC BY 4.0. https://tokenbarometer.com/compare/deepseek_deepseek-chat-vs-meta_llama-3.3-70b-instruct

DeepSeek V3.2 (chat) page | Llama 3.3 70B Instruct page | Open DeepSeek V3.2 (chat) in calculator | Open Llama 3.3 70B Instruct in calculator