Llama-3.3-70B-Instruct

The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out). The Llama 3.3 instruction tuned text only model is optimized for multilingual dialogue use cases and outperforms many of the available open source and closed chat models on common industry benchmarks.

Supported languages: English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai.

Model Card

meta/llama-3.3-70b-instruct

Organization: Meta
Family: llama
Providers: 40
Context: 128,000
Output limit: 4,096
Knowledge: 2023-12
Release: 2024-12-06
Updated: —
Weights: Open
Input: text
Output: text
Capabilities: —

Providers

Compare every cataloged offer for this model, including its provider, route, token limits, price, and capabilities.

59 offers · USD per million tokens

Provider	Model ID	Context	Output	Input / output · 1M	Reasoning	Tools	Structured	Details
Abacus	llama-3.3-70b-versatile	128,000	32,768	— / —	—	—	—	View
Abacus	meta-llama/Meta-Llama-3.3-70B-Instruct	131,072	8,192	— / —	—	—	—	View
AI/ML API	meta-llama/Llama-3.3-70B-Instruct-Turbo	128,000	127,000	— / —	—	—	—	View
Amazon Bedrock	meta.llama3-3-70b-instruct-v1:0	128,000	4,096	— / —	—	—	—	View
Amazon Bedrock	us.meta.llama3-3-70b-instruct-v1:0	128,000	4,096	— / —	—	—	—	View
CloudFerro Sherlock	meta-llama/Llama-3.3-70B-Instruct	70,000	70,000	— / —	—	—	—	View
Cloudflare Workers AI	@cf/meta/llama-3.3-70b-instruct-fp8-fast	24,000	24,000	— / —	—	—	—	View
CoreWeave	meta-llama/Llama-3.3-70B-Instruct	128,000	128,000	— / —	—	—	—	View
Cortecs	llama-3.3-70b-instruct	131,000	131,000	— / —	—	—	—	View
Crusoe	meta-llama/Llama-3.3-70B-Instruct	128,000	4,096	— / —	—	—	—	View
Deep Infra	meta-llama/Llama-3.3-70B-Instruct-Turbo	131,072	16,384	— / —	—	—	—	View
DevPass (LLM Gateway)	llama-3.3-70b-instruct	131,072	4,096	— / —	—	—	—	View
Eden AI	scaleway/llama-3.3-70b-instruct	128,000	4,096	— / —	—	—	—	View
Eden AI	arkanecloud/meta-llama/Llama-3.3-70B-Instruct	128,000	4,096	— / —	—	—	—	View
Eden AI	deepinfra/meta-llama/Llama-3.3-70B-Instruct	131,072	4,096	— / —	—	—	—	View
Eden AI	greenpt/llama-3.3-70b-instruct	128,000	4,096	— / —	—	—	—	View
Eden AI	ionos/meta-llama/Llama-3.3-70B-Instruct	128,000	4,096	— / —	—	—	—	View
evroc	nvidia/Llama-3.3-70B-Instruct-FP8	128,000	4,096	— / —	—	—	—	View
GreenPT	llama-3.3-70b-instruct	100,000	16,384	— / —	—	—	—	View
Groq	llama-3.3-70b-versatile	131,072	32,768	— / —	—	—	—	View
Helicone	llama-3.3-70b-versatile	131,072	32,678	— / —	—	—	—	View
Helicone	llama-3.3-70b-instruct	128,000	16,400	— / —	—	—	—	View
Hugging Face	meta-llama/Llama-3.3-70B-Instruct	131,072	4,096	— / —	—	—	—	View
IO.NET	meta-llama/Llama-3.3-70B-Instruct	128,000	4,096	— / —	—	—	—	View
Kilo Gateway	meta-llama/llama-3.3-70b-instruct	131,072	16,384	— / —	—	—	—	View
LLM Gateway	cerebras/llama-3.3-70b-instruct	128,000	4,096	— / —	—	—	—	View
LLM Gateway	novita/llama-3.3-70b-instruct	131,072	120,000	— / —	—	—	—	View
Meganova	meta-llama/Llama-3.3-70B-Instruct	131,072	16,384	— / —	—	—	—	View
Merge Gateway	meta/llama-3.3-70b-instruct	131,072	32,768	— / —	—	—	—	View
NanoGPT	meta-llama/llama-3.3-70b-instruct	131,072	16,384	— / —	—	—	—	View
NanoGPT	TEE/llama3-3-70b	128,000	16,384	— / —	—	—	—	View
Neon	meta-llama-3-3-70b-instruct	128,000	8,192	— / —	—	—	—	View
Novita AI	meta-llama/llama-3.3-70b-instruct	12,288	12,288	— / —	—	—	—	View
OCI Generative AI	meta.llama-3.3-70b-instruct	128,000	4,096	— / —	—	—	—	View
AkashMLvia OpenRouter	meta-llama/llama-3.3-70b-instruct fp8	131,072	128,000	$0.20 / $0.52	No	Yes	Yes	View
Cloudflarevia OpenRouter	meta-llama/llama-3.3-70b-instruct fp8	24,000	21,600	$0.293 / $2.253	No	No	No	View
CoreWeavevia OpenRouter	meta-llama/llama-3.3-70b-instruct fp16	128,000	115,200	$0.71 / $0.71	No	Yes	Yes	View
DeepInfravia OpenRouter	meta-llama/llama-3.3-70b-instruct fp8	131,072	16,384	$0.10 / $0.32	No	Yes	Yes	View
Googlevia OpenRouter	meta-llama/llama-3.3-70b-instruct unknown	128,000	8,192	$0.72 / $0.72	No	Yes	Yes	View
Googlevia OpenRouter	meta-llama/llama-3.3-70b-instruct unknown	128,000	115,200	$0.72 / $0.72	No	No	No	View
Groqvia OpenRouter	meta-llama/llama-3.3-70b-instruct unknown	131,072	32,768	$0.59 / $0.79	No	Yes	No	View
Novitavia OpenRouter	meta-llama/llama-3.3-70b-instruct bf16	12,288	11,059	$0.135 / $0.40	No	Yes	No	View
Parasailvia OpenRouter	meta-llama/llama-3.3-70b-instruct fp8	131,072	16,384	$0.22 / $0.50	No	No	Yes	View
SambaNovavia OpenRouter	meta-llama/llama-3.3-70b-instruct unknown	131,072	3,072	$0.45 / $0.90	No	No	Yes	View
Togethervia OpenRouter	meta-llama/llama-3.3-70b-instruct unknown	131,072	2,048	$1.04 / $1.04	No	Yes	Yes	View
OVHcloud AI Endpoints	meta-llama-3_3-70b-instruct	131,072	131,072	— / —	—	—	—	View
Pendra	llama3.3:70b	128,000	4,096	— / —	—	—	—	View
Poe	cerebras/llama-3.3-70b-cs	0	0	— / —	—	—	—	View
Prime Inference	meta-llama/llama-3.3-70b-instruct	131,072	16,384	— / —	—	—	—	View
Regolo AI	llama-3.3-70b-instruct	128,000	16,384	— / —	—	—	—	View
SambaNova	Meta-Llama-3.3-70B-Instruct	131,072	3,072	— / —	—	—	—	View
Scaleway	llama-3.3-70b-instruct	100,000	16,384	— / —	—	—	—	View
Snowflake Cortex	snowflake-llama3.3-70b	128,000	4,096	— / —	—	—	—	View
STACKIT	cortecs/Llama-3.3-70B-Instruct-FP8-Dynamic	128,000	4,096	— / —	—	—	—	View
Tinfoil	llama3-3-70b	131,072	4,096	— / —	—	—	—	View
Together AI	meta-llama/Llama-3.3-70B-Instruct-Turbo	131,072	131,072	— / —	—	—	—	View
Venice AI	llama-3.3-70b	128,000	4,096	— / —	—	—	—	View
Vercel AI Gateway	meta/llama-3.3-70b	128,000	8,192	— / —	—	—	—	View
watsonx.ai	meta-llama/llama-3-3-70b-instruct	131,072	4,096	— / —	—	—	—	View

Prices and limits apply to each offer. A dash means the value is not provided. OpenRouter offers show the hosting provider and routing channel separately.

Pricing

Compare token rates and cache charges across available routes and hosting configurations.

USD per million tokens

Input from$0.10Per 1M input tokens

Output from$0.32Per 1M output tokens

Available offers59Across 40 providers

Lowest listed input and output prices across the offers above; they may belong to different providers. These are base rates, before conditional pricing, discounts or additional fees.

Endpoint price comparisonLatest observed snapshot · 10 lowest-priced configurations

InputOutput

DeepInfradeepinfra/turbo

$0.10$0.32

Novitanovita/bf16

$0.135$0.40

AkashMLakashml/fp8

$0.20$0.52

Parasailparasail/fp8

$0.22$0.50

Cloudflarecloudflare/fp8

$0.293$2.253

SambaNovasambanova-turbo

$0.45$0.90

Groqgroq

$0.59$0.79

CoreWeavecoreweave/fp16

$0.71$0.71

Googlegoogle-vertex/us-central1

$0.72$0.72

Googlegoogle-vertex

$0.72$0.72

meta-llama/llama-3.3-70b-instructObserved Oct 10, 2026, 11:31 AM UTC

Input: $0.22
Output: $0.50
Cache read: $0.11

Hosting prices through OpenRouter All endpoints and cache rates

Endpoint	Input / 1M	Output / 1M	Cache read / 1M	Cache write / 1M
AkashMLakashml/fp8	$0.20	$0.52	$0.10	—
Cloudflarecloudflare/fp8	$0.293	$2.253	—	—
CoreWeavecoreweave/fp16	$0.71	$0.71	$0.71	—
DeepInfradeepinfra/turbo	$0.10	$0.32	—	—
Googlegoogle-vertex/us-central1	$0.72	$0.72	—	—
Googlegoogle-vertex	$0.72	$0.72	—	—
Groqgroq	$0.59	$0.79	$0.295	—
Novitanovita/bf16	$0.135	$0.40	—	—
Parasailparasail/fp8	$0.22	$0.50	$0.11	—
SambaNovasambanova-turbo	$0.45	$0.90	—	—
Togethertogether	$1.04	$1.04	—	—

Each row is a hosting configuration. Open an offer in Providers for its full pricing conditions and additional fees. Endpoint observations may differ from the routing catalog's base price.

Performance

Review reported latency and output-throughput percentiles from the latest observed rolling window.

Latest 30-minute observation

Time to first token · milliseconds

Latency distributionRolling 30-minute snapshot · 6 endpoints

P50P75P90P99

AkashMLakashml/fp8

544–4,125.8 ms

Cloudflarecloudflare/fp8

454–4,511.5 ms

CoreWeavecoreweave/fp16

814–1,793.88 ms

DeepInfradeepinfra/turbo

475–2,635.18 ms

Googlegoogle-vertex/us-central1

1,495–1,495 ms

Groqgroq

348–1,217.54 ms

All latency percentiles 10 endpoints

Endpoint	P50	P75	P90	P99
AkashMLakashml/fp8	544 ms	787 ms	1,217 ms	4,125.8 ms
Cloudflarecloudflare/fp8	454 ms	713.5 ms	1,332 ms	4,511.5 ms
CoreWeavecoreweave/fp16	814 ms	1,119 ms	1,364.2 ms	1,793.88 ms
DeepInfradeepinfra/turbo	475 ms	647.25 ms	966.8 ms	2,635.18 ms
Googlegoogle-vertex/us-central1	1,495 ms	1,495 ms	1,495 ms	1,495 ms
Groqgroq	348 ms	494 ms	693 ms	1,217.54 ms
Novitanovita/bf16	935 ms	1,267 ms	1,704 ms	6,514.75 ms
Parasailparasail/fp8	724 ms	1,149 ms	2,435 ms	8,132.4 ms
SambaNovasambanova-turbo	676 ms	811.5 ms	1,126.2 ms	2,388.98 ms
Togethertogether	576 ms	927.5 ms	1,410.4 ms	4,595 ms

Output throughput · tokens per second

Throughput distributionRolling 30-minute snapshot · 6 endpoints

P50P75P90P99

AkashMLakashml/fp8

35–49 t/s

Cloudflarecloudflare/fp8

49–74 t/s

CoreWeavecoreweave/fp16

71–93 t/s

DeepInfradeepinfra/turbo

20–30 t/s

Googlegoogle-vertex/us-central1

119–119 t/s

Groqgroq

236–423.54 t/s

All throughput percentiles 10 endpoints

Endpoint	P50	P75	P90	P99
AkashMLakashml/fp8	35 t/s	40 t/s	44 t/s	49 t/s
Cloudflarecloudflare/fp8	49 t/s	58 t/s	65 t/s	74 t/s
CoreWeavecoreweave/fp16	71 t/s	77 t/s	85 t/s	93 t/s
DeepInfradeepinfra/turbo	20 t/s	23 t/s	25 t/s	30 t/s
Googlegoogle-vertex/us-central1	119 t/s	119 t/s	119 t/s	119 t/s
Groqgroq	236 t/s	292 t/s	343 t/s	423.54 t/s
Novitanovita/bf16	38 t/s	43 t/s	46 t/s	50 t/s
Parasailparasail/fp8	43 t/s	55 t/s	71 t/s	102 t/s
SambaNovasambanova-turbo	162 t/s	222.25 t/s	246 t/s	297.1 t/s
Togethertogether	59 t/s	74 t/s	91.1 t/s	132.63 t/s

Percentiles describe the observed request distribution. These are snapshots of a rolling window, not a historical time series.

Uptime

Compare successful request rates across the latest 5-minute, 30-minute, and 24-hour observations.

Successful requests by window

Availability by endpointCurrent overlapping windows · 11 endpoints

Endpoint5 minutes30 minutes24 hours

AkashMLakashml/fp8Observed Oct 10, 2026, 11:31 AM UTC

100%

Cloudflarecloudflare/fp8Observed Oct 10, 2026, 11:31 AM UTC

100%

99.86%

CoreWeavecoreweave/fp16Observed Oct 10, 2026, 11:31 AM UTC

100%

99.58%

DeepInfradeepinfra/turboObserved Oct 10, 2026, 11:31 AM UTC

100%

94.16%

Googlegoogle-vertex/us-central1Observed Oct 10, 2026, 11:31 AM UTC

—

Googlegoogle-vertexObserved Oct 10, 2026, 11:31 AM UTC

—

GroqgroqObserved Oct 10, 2026, 11:31 AM UTC

100%

Novitanovita/bf16Observed Oct 10, 2026, 11:31 AM UTC

100%

97.24%

Parasailparasail/fp8Observed Oct 10, 2026, 11:31 AM UTC

100%

99.93%

SambaNovasambanova-turboObserved Oct 10, 2026, 11:31 AM UTC

100%

TogethertogetherObserved Oct 10, 2026, 11:31 AM UTC

100%

96.05%

Reported by OpenRouter at the time of collection; rate-limited requests are excluded. A dash means no measurement was reported. This is not a live availability check.

Benchmarks

Review attributed evaluation results on each source's original scale.

External evaluations

meta-llama/llama-3.3-70b-instructObserved Oct 10, 2026, 11:31 AM UTC

Benchmark profileArtificial Analysis · shared source scale from 0 to 20

Coding

11.9

Intelligence

7.7

Sources: Artificial Analysis and Design Arena, via OpenRouter. Scores retain their original scales and are not OpenAI Suite ratings.

Detailed evaluations

Evaluation scoresAccuracy or item F1 · higher is better

gpqa diamondMeta: Llama 3.3 70B Instruct

47.159%

tau bench verified airlineMeta: Llama 3.3 70B Instruct

31.987%

Evaluation	Configuration	Source	Results	Details
gpqa diamond	Meta: Llama 3.3 70B Instructmeta-llama/llama-3.3-70b-instruct	OpenRouter	47.159%	Details
Composite indices	Llama 3.3 Instruct 70Bmeta-llama/llama-3.3-70b-instruct	Artificial Analysis	Intelligence 7.7 · Coding 11.9 · Agentic —	Details
tau bench verified airline	Meta: Llama 3.3 70B Instructmeta-llama/llama-3.3-70b-instruct	OpenRouter	31.987%	Details

Sources: Artificial Analysis, Design Arena and OpenRouter. Evaluation configurations and scales differ; results are not interchangeable. Updated Oct 24, 2026.

Apps & session costs

Compare observed median session costs for applications using this model, grouped by conversation length.

30-day sample · through Oct 18, 2026

Application	Model route	Turns	Median cost / session	Details
Hermes Agent	meta-llama/llama-3.3-70b-instruct	2–9 turns	$0.008046	Details
Hermes Agent	meta-llama/llama-3.3-70b-instruct	10–49 turns	$0.041944	Details
Hermes Agent	meta-llama/llama-3.3-70b-instruct	1 turn	$0.001164	Details

Source: OpenRouter, as of Oct 20, 2026. Published under CC BY 4.0. These are observed costs per session, not token prices or a forecast for your workload. Apps are compared separately.

Technical details & API

Explore specifications, supported parameters, reasoning controls, and API identifiers for each OpenRouter route.

Current route specifications

meta-llama/llama-3.3-70b-instructObserved Oct 10, 2026, 11:31 AM UTC

Context: 131,072
Maximum output: 16,384
Tokenizer: Llama3
Input: text
Output: text
Moderation: No
Instruction format: llama3
Knowledge cutoff: 2023-12-31
Added to OpenRouter: 2024-12-06

Base URLhttps://openrouter.ai/api/v1

Model IDmeta-llama/llama-3.3-70b-instruct

Version IDmeta-llama/llama-3.3-70b-instruct

Weights IDmeta-llama/Llama-3.3-70B-Instruct

Supported parameters

frequency_penalty
logit_bias
logprobs
max_tokens
min_p
presence_penalty
repetition_penalty
response_format
seed
stop
structured_outputs
temperature
tool_choice
tools
top_k
top_logprobs
top_p

Frequently asked questions

Answers based on the model metadata and serving offers currently in the catalog.

How much does Llama-3.3-70B-Instruct cost?

Listed rates start at $0.10 per million input tokens and $0.32 per million output tokens. Prices vary by provider and configuration; see Pricing for details.

Which providers offer Llama-3.3-70B-Instruct?

There are 59 cataloged offers across 40 providers. Compare model IDs, prices and limits in Providers.

What is the context window?

The cataloged model context is 128,000 tokens. Each serving endpoint may apply a different limit.

Does it support tools and structured output?

Tool calling: Not reported. Structured output: Not reported. Support can vary by endpoint.

Resources

Hugging Face ↗

Catalog history · 3 recorded revisions

Metadata observations since this model was first cataloged. These are not model release versions.

Oct 20, 2026 · 02ae76fa284c
Sep 14, 2026 · cafdb74fee26
Sep 12, 2026 · 8e4e79035892