Cheapest Llama 3.3 70B Instruct input price

The Cheapest Llama 3.3 70B Instruct input price is USD 0.10 per 1M input tokens on 2026-10-24 (n=9 hosts), set by inference/openrouter/meta-llama/llama-3.3-70b-instruct/deepinfra/fp8. Definition: The lowest input price, in USD per million tokens, for Llama 3.3 70B Instruct among the hosts listing it (OpenRouter endpoints and first-party host price pages), with the host and quantization of the cheapest endpoint stated.

Computed from the SKU state at the end of each UTC day (latest value: 2026-10-24; today's value is recomputed until the day ends). The constituents of the latest value are listed below; each shows when it was verified and links to its source.

Change over this window: 0.0% from 2026-10-19 to 2026-10-24 (like for like, on the constituents both days share).

Dashed lines with a diamond mark days on which the set of constituents (n) changed; a jump on such a day can be coverage, not a price move.

Download this chart: SVG · PNG · CSV · JSON

Constituents of the latest value

Details of the latest value

Model

meta-llama/llama-3.3-70b-instruct

Name

Llama 3.3 70B Instruct

Min hosts

3

Min coverage

3

First party skus

33

Openrouter rows replaced by first party

21

Community rows excluded

177

Hosts

9

SKU ID

inference/openrouter/meta-llama/llama-3.3-70b-instruct/deepinfra/fp8

Host

DeepInfra

Quantization

fp8

output price (USD per 1M)

0.32

median across hosts (USD per 1M)

0.293

Basis

cheapest endpoint over all listed quantizations; its quantization is stated above, the cheapest per quantization is in by_quantization

stale constituents left out

0

coverage changed

yes

{
 "by_quantization": {
  "bf16": {
   "host": "Novita",
   "hosts": 1,
   "min_usd_per_1m": 0.135
  },
  "fp16": {
   "host": "CoreWeave",
   "hosts": 1,
   "min_usd_per_1m": 0.71
  },
  "fp8": {
   "host": "DeepInfra",
   "hosts": 4,
   "min_usd_per_1m": 0.1
  },
  "unknown": {
   "host": "Groq",
   "hosts": 3,
   "min_usd_per_1m": 0.59
  }
 },
 "chain": {
  "broken": false,
  "change_pct": 0.0,
  "index": 100.0,
  "matched": 9,
  "members_entered": 0,
  "members_left": 1,
  "prev_day": "2026-10-23",
  "ratio": 1.0
 },
 "constituents": {
  "akashml|inference/openrouter/meta-llama/llama-3.3-70b-instruct/akashml/fp8": 0.2,
  "cloudflare|inference/openrouter/meta-llama/llama-3.3-70b-instruct/cloudflare/fp8": 0.293,
  "coreweave|inference/openrouter/meta-llama/llama-3.3-70b-instruct/coreweave/fp16": 0.71,
  "deepinfra|inference/openrouter/meta-llama/llama-3.3-70b-instruct/deepinfra/fp8": 0.1,
  "google|inference/openrouter/meta-llama/llama-3.3-70b-instruct/google/unknown": 0.72,
  "groq|inference/openrouter/meta-llama/llama-3.3-70b-instruct/groq/unknown": 0.59,
  "novita|inference/openrouter/meta-llama/llama-3.3-70b-instruct/novita/bf16": 0.135,
  "parasail|inference/openrouter/meta-llama/llama-3.3-70b-instruct/parasail/fp8": 0.22,
  "sambanova|inference/sambanova/meta-llama/llama-3.3-70b-instruct/sambanova/unknown": 0.6
 },
 "first_party_hosts": [
  "BaseTen",
  "Cerebras",
  "Groq",
  "SambaNova",
  "Together"
 ],
 "left": [
  "together"
 ]
}

Definition

The lowest input price, in USD per million tokens, for Llama 3.3 70B Instruct among the hosts listing it (OpenRouter endpoints and first-party host price pages), with the host and quantization of the cheapest endpoint stated.

Inclusion rule

inference rows of the model meta-llama/llama-3.3-70b-instruct (exact OpenRouter id; :batch, :free and other variants are separate models and not included): OpenRouter endpoints plus first-party host SKUs (groq, together, sambanova, cerebras, baseten), which publish no quantization and so are 'unknown'; a host with a first-party SKU for the model counts once, with its first-party price (its OpenRouter endpoints are dropped); community rows (hf-router) never enter; the model must be open-weight per llm/modelsdev (the same join and grouping as open-model-host-spread); endpoints in a healthy status and default service tier; price 0 (free) dropped; the headline is the cheapest endpoint over all quantizations (its quantization is in the metadata, the cheapest per quantization in by_quantization). Not computed below 3 distinct hosts. List prices of the host, not a transacted price.

When it cannot be computed: Not computed: that day shows no number and states the reason; the last full value stays shown with its date. A source whose last confirmed snapshot is older than 3x its cadence (6h sources 18 h, 12h sources 36 h, 1d sources 72 h) drops out; below minimum coverage the series is not computed and says 'not verified: <source> stale since <date>'.

Minimum coverage: 3 hosts. Below that the series is not computed and says so instead of showing a number.

Sources feeding it

openrouter-endpoints, modelsdev, groq-models, together-pricing, sambanova-models, cerebras-models, baseten-pricing

Deliberately excluded

• free variants (price 0)
• unhealthy or non-default-tier endpoints
• batch and other model variants
• hf-inference-providers (community cross-check)

What can break it

• the cheapest host may run a lower-precision quantization (stated in the metadata)
• promotions and temporary discounts
• a model changing its OpenRouter id
• hosts that list but do not serve
• OpenRouter changing its endpoint schema

Daily values, last 90 days

Days with a value since 2026-10-19 (days without one are left out of the data files): CSV · JSON.

API: JSON · CSV

InfraRabbit (2026). Cheapest Llama 3.3 70B Instruct input price [data as of 2026-10-24]. https://infrarabbit.com/index/inference-cheapest-llama-3-3-70b (accessed 2026-10-24). List prices, not transacted prices. CC BY 4.0.

@misc{infrarabbit_index_inference_cheapest_llama_3_3_70b_2026,
  author       = {{InfraRabbit}},
  title        = {Cheapest Llama 3.3 70B Instruct input price},
  year         = {2026},
  howpublished = {\url{https://infrarabbit.com/index/inference-cheapest-llama-3-3-70b}},
  note         = {Data as of 2026-10-24; accessed 2026-10-24; list prices, not transacted prices; CC BY 4.0}
}