BroadRouter

Llama 3.1 70B vs Llama 3.1 8B Instruct API pricing

Updated

Llama 3.1 70B: USD 0.72 input, USD 0.72 output per 1M tokens (Amazon Bedrock). Llama 3.1 8B Instruct: USD 0.02 input, USD 0.04 output per 1M tokens (DeepInfra). At 3 input tokens per output token, Llama 3.1 70B costs 29× as much as Llama 3.1 8B Instruct (USD 0.72 vs USD 0.025 per 1M).

Llama 3.1 70B and Llama 3.1 8B Instruct are neighbouring sizes of the same Meta generation (70B and 8B parameters, by name). Price only: every number is computed from prices read from each provider's own pricing page (linked per cell, newest check 2026-11-15), with no claim about quality, speed or accuracy.

How do Llama 3.1 70B and Llama 3.1 8B Instruct differ on price?

Llama 3.1 70B and Llama 3.1 8B Instruct 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.

Llama 3.1 70B and Llama 3.1 8B Instruct API prices compared
PriceLlama 3.1 70BLlama 3.1 8B Instruct
Model pageLlama 3.1 Instruct (70B)
Meta · 3 providers
Llama 3.1 8B Instruct
Meta · 6 providers
Cheapest standard price, blended 3:1USD 0.72
USD 0.72 in / USD 0.72 out at Amazon Bedrock
Amazon Bedrock page checked 2026-11-14
USD 0.025 lower
USD 0.02 in / USD 0.04 out at DeepInfra
DeepInfra page checked 2026-11-14
Cheapest inputUSD 0.293
Cloudflare Workers AI source
USD 0.02 lower
DeepInfra source
Cheapest cached inputNo standard price lists a cache price; cached tokens bill at the input price.USD 0.028
Featherless source
Cheapest outputUSD 0.72
Amazon Bedrock source
USD 0.04 lower
DeepInfra source
Maker's own APIMeta 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 priceUSD 0.72 / USD 2.253
input / output, 3 prices
USD 0.21 / USD 0.22
input / output, 6 prices
Context window32,768 tokens, largest stated131,072 tokens, largest stated; from 16,384 at some providers
Providers with a standard price36

Report a wrong price or number (email to info@broadnet.ai, with this page's link filled in).

What do Llama 3.1 70B and Llama 3.1 8B Instruct 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).

Monthly cost of three workloads on Llama 3.1 70B and Llama 3.1 8B Instruct
WorkloadLlama 3.1 70BLlama 3.1 8B InstructDifference
Support chatbot
2,000 requests a day, 1,500 input tokens (60% cached), 300 output
USD 66.92
Cloudflare Workers AI
USD 2.52
DeepInfra
USD 64.40
Llama 3.1 8B Instruct lower
RAG over documents
1,000 requests a day, 8,000 input tokens (30% cached), 500 output
USD 104.11
Cloudflare Workers AI
USD 5.40
DeepInfra
USD 98.71
Llama 3.1 8B Instruct lower
Coding agent (one developer)
1,500 requests a day, 40,000 input tokens (85% cached), 2,000 output incl. reasoning
USD 730.17
Cloudflare Workers AI
USD 39.60
DeepInfra
USD 690.57
Llama 3.1 8B Instruct lower

Llama 3.1 8B Instruct is cheaper on all three workloads. The widest gap is on Support chatbot: the dearer model costs 27× as much as the cheaper one.

Does the cheaper model change with your input:output ratio?

Llama 3.1 8B Instruct is cheaper at every mix of input and output tokens, before cache discounts: its cheapest input price and its cheapest output price are both at or below Llama 3.1 70B's, so the input:output ratio never changes the answer.

Price per 1M tokens of Llama 3.1 70B and Llama 3.1 8B Instruct at different input:output mixes
Input : output tokensLlama 3.1 70BLlama 3.1 8B Instruct
1 : 3 (long answers)USD 0.72
Amazon Bedrock
USD 0.035 lower
DeepInfra
1 : 1USD 0.72
Amazon Bedrock
USD 0.03 lower
DeepInfra
3 : 1 (the site’s blend)USD 0.72
Amazon Bedrock
USD 0.025 lower
DeepInfra
10 : 1 (retrieval, long prompts)USD 0.4712
Cloudflare Workers AI
USD 0.02182 lower
DeepInfra
30 : 1 (agents, big contexts)USD 0.3562
Cloudflare Workers AI
USD 0.02065 lower
DeepInfra

USD per 1M tokens at the cheapest standard provider for each mix, before cache discounts. Thinking tokens bill as output.

What do Llama 3.1 70B and Llama 3.1 8B Instruct cost per character in other languages?

Cost to send 1M characters at each model's cheapest non-promotional standard input price, with the Llama 3.3 70B tokenizer for Llama 3.1 70B (the same one for both, so the gap equals the price gap):

Input cost per 1M characters by language, Llama 3.1 70B and Llama 3.1 8B Instruct
ContentLlama 3.1 70BLlama 3.1 8B Instruct
EnglishUSD 0.064
Cloudflare Workers AI
USD 0.0043 lower
DeepInfra
Arabic MSAUSD 0.113
Cloudflare Workers AI
USD 0.0077 lower
DeepInfra
Egyptian ArabicUSD 0.115
Cloudflare Workers AI
USD 0.0079 lower
DeepInfra
HindiUSD 0.153
Cloudflare Workers AI
USD 0.010 lower
DeepInfra
ChineseUSD 0.234
Cloudflare Workers AI
USD 0.016 lower
DeepInfra
SpanishUSD 0.078
Cloudflare Workers AI
USD 0.0053 lower
DeepInfra
CodeUSD 0.075
Cloudflare Workers AI
USD 0.0051 lower
DeepInfra

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

More comparisons: Llama 3.1 70B vs Llama 3 70B · Llama 3.3 70B vs Llama 3.1 70B · Llama 3.1 8B Instruct vs Llama 3 8B · every Llama 3.1 70B comparison · every Llama 3.1 8B Instruct comparison · all 662 price comparisons

Questions

Is Llama 3.1 70B cheaper than Llama 3.1 8B Instruct?

No. On a blend of 3 input tokens to 1 output token, Llama 3.1 8B Instruct costs USD 0.025 per 1M tokens and Llama 3.1 70B USD 0.72, at each model's cheapest standard price. For input alone, Llama 3.1 8B Instruct is cheaper (USD 0.02 against USD 0.293 per 1M); for output alone, Llama 3.1 8B Instruct is (USD 0.04 against USD 0.72). This compares price only, not quality.

What do Llama 3.1 70B and Llama 3.1 8B Instruct cost per 1M tokens?

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. Llama 3.1 8B Instruct: from USD 0.02 input and USD 0.04 output at DeepInfra; median of 6 standard prices USD 0.21 / USD 0.22. Each price is read from the provider's own pricing page; the source is linked in the table.

Which providers sell both Llama 3.1 70B and Llama 3.1 8B Instruct?

3 providers have a standard price for both: Amazon Bedrock (USD 0.72 and USD 0.22 at 3:1), Cloudflare Workers AI (USD 0.783 and USD 0.4182 at 3:1), Featherless (USD 2.88 and USD 0.23 at 3:1). The first is the lowest combined price. Buying each model from its own cheapest provider costs USD 0.72 and USD 0.025.

Which is cheaper for a coding agent or a chatbot, Llama 3.1 70B or Llama 3.1 8B Instruct?

For a coding agent (1,500 requests a day, 40,000 input tokens (85% cached), 2,000 output incl. reasoning): USD 730.17 a month on Llama 3.1 70B (Cloudflare Workers AI) against USD 39.60 on Llama 3.1 8B Instruct (DeepInfra), a difference of USD 690.57. For a support chatbot (2,000 requests a day, 1,500 input tokens (60% cached), 300 output): USD 66.92 a month on Llama 3.1 70B (Cloudflare Workers AI) against USD 2.52 on Llama 3.1 8B Instruct (DeepInfra), a difference of USD 64.40. 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 Llama 3.1 70B and Llama 3.1 8B Instruct change?

Llama 3.1 8B Instruct is cheaper at every mix of input and output tokens, before cache discounts: its cheapest input price and its cheapest output price are both at or below Llama 3.1 70B's, so the input:output ratio never changes the answer.

Do Llama 3.1 70B and Llama 3.1 8B Instruct have cached input pricing?

Llama 3.1 70B: no standard price lists a cached-input discount. Llama 3.1 8B Instruct from USD 0.028 per 1M cached input tokens (Featherless; 1 of 6 standard prices list a discount). Cached tokens are billed at the input price where no cache price is published.

Which has the larger context window, Llama 3.1 70B or Llama 3.1 8B Instruct?

Llama 3.1 8B Instruct: up to 131,072 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.

Read as MarkdownAsk ChatGPTAsk ClaudeAsk Perplexity