DeepSeek R1 0528 vs Llama 3.1 70B API pricing
Updated
DeepSeek R1 0528: USD 0.385 input, USD 1.625 output per 1M tokens (Featherless). Llama 3.1 70B: USD 0.72 input, USD 0.72 output per 1M tokens (Amazon Bedrock). At 3 input tokens per output token, Llama 3.1 70B costs 3.6% more than DeepSeek R1 0528 (USD 0.72 vs USD 0.695 per 1M).
How do DeepSeek R1 0528 and Llama 3.1 70B differ on price?
- Cheapest input price: Llama 3.1 70B USD 0.293 against USD 0.385 per 1M tokens, so DeepSeek R1 0528 costs 1.31× as much (31% more).
- Cheapest output price: Llama 3.1 70B USD 0.72 against USD 1.625 per 1M tokens, so DeepSeek R1 0528 costs 2.3× as much.
- Among the 201 indexed models, DeepSeek R1 0528 ranks 110th cheapest and Llama 3.1 70B 113th on the 3:1 blend.
- The cheaper model changes at 2.7 input tokens per output token and 6.8 input tokens per output token.
- Only DeepSeek R1 0528 lists a cached-input discount: DeepSeek R1 0528 at 3 standard prices of 5, from USD 0.23 (GMI Cloud, 60% below that provider's input price). Llama 3.1 70B's cached tokens bill at the input price.
- DeepSeek R1 0528 states up to 163,840 tokens of context, 5× the 32,768 of Llama 3.1 70B.
- DeepSeek R1 0528 has 5 providers with a standard price, 2 more than Llama 3.1 70B; 1 provider sells both (Featherless).
- On one account, Featherless charges USD 0.695 for DeepSeek R1 0528 (its cheapest price) and USD 2.88 for Llama 3.1 70B (4× its cheapest price) at 3:1, the lowest combined price among the providers that sell both.
- Shopping around matters more for Llama 3.1 70B: its dearest standard price is 4× its cheapest, against 3.4× for DeepSeek R1 0528.
- DeepSeek R1 0528 has a batch price at 1 provider, from USD 1.181 at 3:1 (Google Vertex AI); Llama 3.1 70B has a batch price at 1 provider, from USD 0.36 at 3:1 (Amazon Bedrock).
- For the one-developer coding agent preset, the gap is USD 219.24 a year in DeepSeek R1 0528's favour (DeepSeek R1 0528 USD 711.90 a month, Llama 3.1 70B USD 730.17).
- Per character of input, the gap depends on the language: DeepSeek R1 0528 costs 1.45× as much as Llama 3.1 70B on Hindi text but 4% less than Llama 3.1 70B on Chinese, because the two tokenizers split those languages differently.
DeepSeek R1 0528 and Llama 3.1 70B prices side by side
USD per 1M tokens, standard tier. "Cheapest" is the lowest price at 3 input tokens to 1 output token; each line can come from a different provider.
| Price | DeepSeek R1 0528 | Llama 3.1 70B |
|---|---|---|
| Model page | DeepSeek R1 0528 DeepSeek · 5 providers | Llama 3.1 Instruct (70B) Meta · 3 providers |
| Cheapest standard price, blended 3:1 | USD 0.695 lower USD 0.385 in / USD 1.625 out at Featherless Featherless page checked 2026-11-15 | USD 0.72 USD 0.72 in / USD 0.72 out at Amazon Bedrock Amazon Bedrock page checked 2026-11-14 |
| Cheapest input | USD 0.385 Featherless source | USD 0.293 lower Cloudflare Workers AI source |
| Cheapest cached input | USD 0.23 GMI Cloud source | No standard price lists a cache price; cached tokens bill at the input price. |
| Cheapest output | USD 1.625 Featherless source | USD 0.72 lower Amazon Bedrock source |
| Maker's own API | DeepSeek publishes no paid standard price on its own API that we found. | Meta publishes no paid standard price on its own API that we found. |
| Median standard price | USD 0.57 / USD 2.29 input / output, 5 prices | USD 0.72 / USD 2.253 input / output, 3 prices |
| Context window | 163,840 tokens, largest stated; from 131,072 at some providers | 32,768 tokens, largest stated |
| Providers with a standard price | 5 | 3 |
- Lower marks the lower price on a line; tie, equal prices. Other tiers and free rows are on the model pages.
Report a wrong price or number (email to info@broadnet.ai, with this page's link filled in).
What do DeepSeek R1 0528 and Llama 3.1 70B cost for a real workload?
30 days at each model's cheapest standard price for the workload (presets of the cost calculator; cached input at the cached price where listed).
| Workload | DeepSeek R1 0528 | Llama 3.1 70B | Difference |
|---|---|---|---|
| Support chatbot 2,000 requests a day, 1,500 input tokens (60% cached), 300 output | USD 63.90 Featherless | USD 66.92 Cloudflare Workers AI | USD 3.02 DeepSeek R1 0528 lower |
| RAG over documents 1,000 requests a day, 8,000 input tokens (30% cached), 500 output | USD 116.78 Featherless | USD 104.11 Cloudflare Workers AI | USD 12.66 Llama 3.1 70B lower |
| Coding agent (one developer) 1,500 requests a day, 40,000 input tokens (85% cached), 2,000 output incl. reasoning | USD 711.90 GMI Cloud | USD 730.17 Cloudflare Workers AI | USD 18.27 DeepSeek R1 0528 lower |
The cheaper model depends on the workload: DeepSeek R1 0528 for Support chatbot, Llama 3.1 70B for RAG over documents, DeepSeek R1 0528 for Coding agent (one developer). Cache-heavy work rewards a low cached-input price. The widest gap is on RAG over documents: the dearer model costs 12% more than the cheaper one.
Does the cheaper model change with your input:output ratio?
The cheaper model changes with the mix: below 2.7 input tokens per output token, Llama 3.1 70B is cheaper; from 2.7 to 6.8, DeepSeek R1 0528 is cheaper; above 6.8, Llama 3.1 70B is cheaper. At the site's 3:1 blend, DeepSeek R1 0528 is cheaper. More input per output token, as with retrieval or agents that resend long contexts, favours the lower input price; long answers favour the lower output price.
| Input : output tokens | DeepSeek R1 0528 | Llama 3.1 70B |
|---|---|---|
| 1 : 3 (long answers) | USD 1.315 Featherless | USD 0.72 lower Amazon Bedrock |
| 1 : 1 | USD 1.005 Featherless | USD 0.72 lower Amazon Bedrock |
| 3 : 1 (the site’s blend) | USD 0.695 lower Featherless | USD 0.72 Amazon Bedrock |
| 10 : 1 (retrieval, long prompts) | USD 0.4977 Featherless | USD 0.4712 lower Cloudflare Workers AI |
| 30 : 1 (agents, big contexts) | USD 0.425 Featherless | USD 0.3562 lower Cloudflare Workers AI |
USD per 1M tokens at the cheapest standard provider for each mix, before cache discounts. Thinking tokens bill as output.
What do DeepSeek R1 0528 and Llama 3.1 70B cost per character in other languages?
Cost to send 1M characters at each model's cheapest non-promotional standard input price, with the DeepSeek V4.1 Flash tokenizer for DeepSeek R1 0528 and the Llama 3.3 70B tokenizer for Llama 3.1 70B:
| Content | DeepSeek R1 0528 | Llama 3.1 70B |
|---|---|---|
| English | USD 0.082 Featherless | USD 0.064 lower Cloudflare Workers AI |
| Arabic MSA | USD 0.145 Featherless | USD 0.113 lower Cloudflare Workers AI |
| Egyptian Arabic | USD 0.149 Featherless | USD 0.115 lower Cloudflare Workers AI |
| Hindi | USD 0.222 Featherless | USD 0.153 lower Cloudflare Workers AI |
| Chinese | USD 0.226 lower Featherless | USD 0.234 Cloudflare Workers AI |
| Spanish | USD 0.099 Featherless | USD 0.078 lower Cloudflare Workers AI |
| Code | USD 0.102 Featherless | USD 0.075 lower Cloudflare Workers AI |
From the real cost by language dataset (measured tokens per character on FLORES-200 and other texts). Your own text will differ.
Check it on your own traffic
- DeepSeek R1 0528 API pricing: all 5 providers and other tiers.
- Llama 3.1 Instruct (70B) API pricing: all 3 providers and other tiers.
- Cost calculator with DeepSeek R1 0528 and Llama 3.1 70B preloaded
- DeepSeek token counter
- Llama 3 token counter
- Prompt caching calculator
More comparisons: DeepSeek R1 0528 vs DeepSeek R1 · Llama 3.1 70B vs Llama 3 70B · Llama 3.3 70B vs Llama 3.1 70B · Llama 3.1 70B vs Llama 3.1 8B Instruct · every DeepSeek R1 0528 comparison · every Llama 3.1 70B comparison · all 662 price comparisons
Questions
Is DeepSeek R1 0528 cheaper than Llama 3.1 70B?
Yes. On a blend of 3 input tokens to 1 output token, DeepSeek R1 0528 costs USD 0.695 per 1M tokens and Llama 3.1 70B USD 0.72, at each model's cheapest standard price. For input alone, Llama 3.1 70B is cheaper (USD 0.293 against USD 0.385 per 1M); for output alone, Llama 3.1 70B is (USD 0.72 against USD 1.625). This compares price only, not quality.
What do DeepSeek R1 0528 and Llama 3.1 70B cost per 1M tokens?
DeepSeek R1 0528: from USD 0.385 input and USD 1.625 output at Featherless; median of 5 standard prices USD 0.57 / USD 2.29. Llama 3.1 70B: from USD 0.72 input and USD 0.72 output at Amazon Bedrock; median of 3 standard prices USD 0.72 / USD 2.253. Each price is read from the provider's own pricing page; the source is linked in the table.
Which providers sell both DeepSeek R1 0528 and Llama 3.1 70B?
1 provider has a standard price for both: Featherless (USD 0.695 and USD 2.88 at 3:1). Buying each model from its own cheapest provider costs USD 0.695 and USD 0.72.
Which is cheaper for a coding agent or a chatbot, DeepSeek R1 0528 or Llama 3.1 70B?
For a coding agent (1,500 requests a day, 40,000 input tokens (85% cached), 2,000 output incl. reasoning): USD 711.90 a month on DeepSeek R1 0528 (GMI Cloud) against USD 730.17 on Llama 3.1 70B (Cloudflare Workers AI), a difference of USD 18.27. For a support chatbot (2,000 requests a day, 1,500 input tokens (60% cached), 300 output): USD 63.90 a month on DeepSeek R1 0528 (Featherless) against USD 66.92 on Llama 3.1 70B (Cloudflare Workers AI), a difference of USD 3.02. Each uses the cheapest standard price for that workload, cached input included. Try your own numbers.
At what input:output ratio does the cheaper of DeepSeek R1 0528 and Llama 3.1 70B change?
The cheaper model changes with the mix: below 2.7 input tokens per output token, Llama 3.1 70B is cheaper; from 2.7 to 6.8, DeepSeek R1 0528 is cheaper; above 6.8, Llama 3.1 70B is cheaper. At the site's 3:1 blend, DeepSeek R1 0528 is cheaper. More input per output token, as with retrieval or agents that resend long contexts, favours the lower input price; long answers favour the lower output price.
Do DeepSeek R1 0528 and Llama 3.1 70B have cached input pricing?
DeepSeek R1 0528 from USD 0.23 per 1M cached input tokens (GMI Cloud; 3 of 5 standard prices list a discount). Llama 3.1 70B: no standard price lists a cached-input discount. Cached tokens are billed at the input price where no cache price is published.
Which has the larger context window, DeepSeek R1 0528 or Llama 3.1 70B?
DeepSeek R1 0528: up to 163,840 tokens, against 32,768 for Llama 3.1 70B, as stated on providers' pricing pages. Some providers offer less than the maximum; check the model pages before you pick the cheapest row for long prompts.