Llama-3.3-70B

Meta · Instruct · Public weights

Demo activity · 24 hours ending 17 Oct 2026 · Activity, latency and historical quotes are simulated.

Listed providers: 3 2 seen in snapshot
Input / 1M tokens: $0.300–$0.360
Output / 1M tokens: $0.600–$0.720
Model auditing: Not audited

Compare provider offers

3 providers

Input and output savings are shown automatically against the same model’s reference tariff. Provider prices and activity are illustrative.

Reference: DeepInfra · meta-llama/Llama-3.3-70B-Instruct-TurboInput $0.10 USD · Output $0.32 USD / 1MManual capture 2026-10-18 · Reference expired; percentages unavailablePublic model page; live inference not checked.Hosting reference · Standard · FP8 Turbo · 131,072 context. Not a creator list price.

Price comparison assumes 1 = $1 USD. Standard uncached text; the initial 3% P2P fee is included. Chain fees, rent and taxes excluded.

Provider	Input / 1M · vs reference	Output / 1M · vs reference	Completed	Delivery	Median TTFT	Snapshot	Audit	Data retention
node.andes	$0.300 No reference	$0.600 No reference	3,275	99.03%	545 ms	Seen	Not audited	Not declared
node.altiplano	$0.330 No reference	$0.660 No reference	3,592	99.75%	571 ms	Seen	Not audited	Not declared
node.coast	$0.360 No reference	$0.720 No reference	3,308	99.16%	519 ms	Not seen	Not audited	Not declared

Differences compare input and output separately; no token mix is assumed. Published tariffs do not verify model configuration, tokenizer, availability or service quality. Activity follows the selected chart period.

Completed requests

10,175

40.93M tokens processed

Settled consumption

$17.52

Gross consumption · simulated

Complete delivery

99.33%

69 failed of 10,244 requests

Median first token

545 ms

P95 967 ms · 288 simulated timings

Requests

Hourly · UTC

Completed and failed requests · 24 hours

10,175 completed69 failed

Settled consumption

Gross inference consumption · 24 hours

Protocol fees $0.53Seller net $16.99

Price history

/ 1M tokens

Illustrative median quotes across listed providers

InputOutput · dashed

Time to first token

Milliseconds

Simulated timing distribution · 24 hours

Median 545 msP95 967 ms

Quote history is a separate visual scenario. Demo consumption uses current catalog rates, not historical quotes.

Declared capabilities

Context window, tools, vision, precision and rate limits have not been supplied for these example offers. Check each provider’s signed declaration before use.

Audit evidence

No signed audit evidence is connected. Public weights alone do not establish eligibility or grant an audit seal. Demo performance does not create a service reputation score.

View model audit coverage

Catalog listings are illustrative. Live payments, service measurements and signed audits are not connected. Data sources