Meta model product
Llama 3.3 70B
3 is a multilingual large language model optimized for dialogue use cases across multiple languages.
Updated Aug 17, 2026. Default version: Llama 3.3 70B Instruct
LLMBoard Score21.4Llama 3.3 70B Instruct
Coverage100%8 benchmark families
Context window128KTokens
Official input priceN/AOfficial price unavailable
Llama 3.3 70B Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Llama 3.3 70B Benchmark Results
Benchmark scores for Llama 3.3 70B Instruct.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkBFCL v2 | Score77.3% | Rank01 | Participants5 | Percentile100.0% | EvidenceC | Evaluated |
| BenchmarkMBPP EvalPlus | Score87.6% | Rank02 | Participants2 | Percentile0.0% | EvidenceC | Evaluated |
| BenchmarkMGSM | Score91.1% | Rank05 | Participants31 | Percentile86.7% | EvidenceC | Evaluated |
| BenchmarkIFEval | Score92.1% | Rank09 | Participants67 | Percentile87.9% | EvidenceC | Evaluated |
| BenchmarkHumanEval | Score88.4% | Rank22 | Participants66 | Percentile67.7% | EvidenceC | Evaluated |
| BenchmarkMATH | Score77.0% | Rank23 | Participants71 | Percentile68.6% | EvidenceC | Evaluated |
| BenchmarkMMLU | Score86.0% | Rank34 | Participants101 | Percentile67.0% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score68.9% | Rank91 | Participants134 | Percentile32.3% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score50.5% | Rank181 | Participants239 | Percentile24.4% | EvidenceC | Evaluated |
Llama 3.3 70B Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenatext style control | Categoryoverall | Rank238 | Rating / score1318.2 | Votes54,778 | ObservationsN/A | Result date |
| Arenatext | Categoryoverall | Rank249 | Rating / score1274.8 | Votes54,778 | ObservationsN/A | Result date |
Llama 3.3 70B Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.05 input, $0.23 output per 1M via NanoGPT
- Tracked offerings
- 37
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderNvidia | Provider model IDmeta/llama-3.3-70b-instruct | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context128K | Updated |
| ProviderLlama | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context128K | Updated |
| ProviderVercel AI Gateway | Provider model IDmeta/llama-3.3-70b | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context128K | Updated |
| ProviderNanoGPT | Provider model IDmeta-llama/llama-3.3-70b-instruct | Regionglobal | Input / 1M$0.05 | Output / 1M$0.23 | Context131.1K | Updated |
| ProviderMeganova | Provider model IDmeta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.10 | Output / 1M$0.30 | Context131.1K | Updated |
| ProviderOpenRouter | Provider model IDmeta-llama/llama-3.3-70b-instruct | Regionglobal | Input / 1M$0.10 | Output / 1M$0.32 | Context131.1K | Updated |
| ProviderEden AI | Provider model IDdeepinfra/meta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.10 | Output / 1M$0.32 | Context131.1K | Updated |
| ProviderKilo Gateway | Provider model IDmeta-llama/llama-3.3-70b-instruct | Regionglobal | Input / 1M$0.10 | Output / 1M$0.32 | Context131.1K | Updated |
| ProviderCortecs | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$0.129 | Output / 1M$0.399 | Context131K | Updated |
| ProviderNebius Token Factory | Provider model IDmeta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.13 | Output / 1M$0.40 | Context128K | Updated |
| ProviderIO.NET | Provider model IDmeta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.13 | Output / 1M$0.38 | Context128K | Updated |
| ProviderHelicone | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$0.13 | Output / 1M$0.39 | Context128K | Updated |
| ProviderEden AI | Provider model IDnebius/meta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.13 | Output / 1M$0.40 | Context131.1K | Updated |
| ProviderLLM Gateway | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$0.13 | Output / 1M$0.40 | Context131.1K | Updated |
| ProviderNovitaAI | Provider model IDmeta-llama/llama-3.3-70b-instruct | Regionglobal | Input / 1M$0.135 | Output / 1M$0.40 | Context131.1K | Updated |
| ProviderMerge Gateway | Provider model IDmeta/llama-3.3-70b-instruct | Regionglobal | Input / 1M$0.22 | Output / 1M$0.50 | Context131.1K | Updated |
| ProviderCrusoe | Provider model IDmeta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.25 | Output / 1M$0.75 | Context128K | Updated |
| ProviderNeon | Provider model IDmeta-llama-3-3-70b-instruct | Regionglobal | Input / 1M$0.50 | Output / 1M$1.5 | Context128K | Updated |
| ProviderHugging Face | Provider model IDmeta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.59 | Output / 1M$0.79 | Context131.1K | Updated |
| ProviderAbacus | Provider model IDmeta-llama/Meta-Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.59 | Output / 1M$0.79 | Context131.1K | Updated |
| ProviderRegolo AI | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$0.60 | Output / 1M$2.7 | Context128K | Updated |
| ProviderCharm Hyper | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$0.608 | Output / 1M$0.838 | Context128K | Updated |
| ProviderAzure Cognitive Services | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$0.71 | Output / 1M$0.71 | Context128K | Updated |
| ProviderAzure | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$0.71 | Output / 1M$0.71 | Context128K | Updated |
| ProviderVertex | Provider model IDmeta/llama-3.3-70b-instruct-maas | Regionglobal | Input / 1M$0.72 | Output / 1M$0.72 | Context128K | Updated |
| ProviderAmazon Bedrock | Provider model IDmeta.llama3-3-70b-instruct-v1:0 | Regionglobal | Input / 1M$0.72 | Output / 1M$0.72 | Context128K | Updated |
| ProviderEden AI | Provider model IDionos/meta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.7519 | Output / 1M$0.7519 | Context128K | Updated |
| Providerwatsonx.ai | Provider model IDmeta-llama/llama-3-3-70b-instruct | Regionglobal | Input / 1M$0.7526 | Output / 1M$0.7526 | Context131.1K | Updated |
| ProviderScaleway | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$0.90 | Output / 1M$0.90 | Context100K | Updated |
| ProviderPioneer | Provider model IDmeta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.90 | Output / 1M$0.90 | Context131.1K | Updated |
| ProviderBerget.AI | Provider model IDmeta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$0.99 | Output / 1M$0.99 | Context128K | Updated |
| ProviderEden AI | Provider model IDscaleway/llama-3.3-70b-instruct | Regionglobal | Input / 1M$1.04 | Output / 1M$1.04 | Context128K | Updated |
| Providerevroc | Provider model IDnvidia/Llama-3.3-70B-Instruct-FP8 | Regionglobal | Input / 1M$1.15 | Output / 1M$1.15 | Context128K | Updated |
| ProviderGreenPT | Provider model IDllama-3.3-70b-instruct | Regionglobal | Input / 1M$1.25 | Output / 1M$1.25 | Context100K | Updated |
| ProviderTinfoil | Provider model IDllama3-3-70b | Regionglobal | Input / 1M$1.75 | Output / 1M$2.75 | Context131.1K | Updated |
| ProviderCloudFerro Sherlock | Provider model IDmeta-llama/Llama-3.3-70B-Instruct | Regionglobal | Input / 1M$2.92 | Output / 1M$2.92 | Context70K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Llama 3.3 70B Runtime Performance
Provider-specific output speed and catalog latency for Llama 3.3 70B Instruct. Runtime does not affect the capability score.
Sort by
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderCerebras | Output Speed2,220 tok/s | Catalog Latency0.65 s | Max Input128K | Max Output128K | Updated |
| ProviderSambanova | Output Speed1,096 tok/s | Catalog Latency0.65 s | Max Input128K | Max Output128K | Updated |
| ProviderGroq | Output Speed268 tok/s | Catalog Latency0.65 s | Max Input128K | Max Output128K | Updated |
| ProviderFireworks | Output Speed197 tok/s | Catalog Latency0.65 s | Max Input128K | Max Output128K | Updated |
| ProviderBedrock | Output Speed100 tok/s | Catalog Latency0.5 s | Max Input128K | Max Output128K | Updated |
| ProviderTogether | Output Speed65 tok/s | Catalog Latency0.65 s | Max Input128K | Max Output128K | Updated |
| ProviderLambda | Output Speed42 tok/s | Catalog Latency0.65 s | Max Input128K | Max Output128K | Updated |
| ProviderHyperbolic | Output Speed42 tok/s | Catalog Latency0.65 s | Max Input128K | Max Output128K | Updated |
| ProviderDeepInfra | Output Speed37 tok/s | Catalog Latency0.65 s | Max Input128K | Max Output128K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Llama 3.3 70B Specifications
Technical details for the model's default version.
- Version
- Llama 3.3 70B Instruct
- Released
- Dec 6, 2024
- Knowledge cutoff
- Dec 1, 2023
- Parameters
- 70B
- Context window
- 128K
- Max output
- 4.1K
- Inputs
- text
- Outputs
- text
- Open weights
- Yes
- License
- Llama 3.3 Community License Agreement
Llama 3.3 70B Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionLlama 3.3 70B Instruct | Released | LLMBoard21.4 | Parameters70B | Context128K | Max output4.1K | Open weightsYes | LicenseLlama 3.3 Community License Agreement |
Llama 3.3 70B vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Llama 3.3 70B
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
ME
Llama 3.1 405B
Meta
ME
Llama 4 Maverick
Meta
ME
Llama 4 Scout
Meta
ME
Llama 3.1 70B
Meta
ME
Llama 3.2 90B
Meta
ME
Llama 3.2 3B
Meta
What is Llama 3.3 70B?
Key information about Llama 3.3 70B and its available data.
3 is a multilingual large language model optimized for dialogue use cases across multiple languages. It is a pretrained and instruction-tuned generative model with 70 billion parameters, outperforming many open-source and closed chat models on common industry benchmarks.
3 supports a context length of 128,000 tokens and is designed for commercial and research use in multiple languages.
Data as of 2026-08-17.
FAQ
Common questions about Llama 3.3 70B.
When was Llama 3.3 70B released?
Llama 3.3 70B's default version was released on Dec 6, 2024.
How much does Llama 3.3 70B cost?
No official standard PAYG price is currently available for Llama 3.3 70B. The lowest tracked third-party offer starts at $0.05 input and $0.23 output via NanoGPT.
Who created Llama 3.3 70B?
Llama 3.3 70B was created by Meta.
What is the context window for Llama 3.3 70B?
The default version has a 128K token context window.
Is Llama 3.3 70B open weight?
Yes. The default version is marked as open weight under Llama 3.3 Community License Agreement.
How many API providers offer Llama 3.3 70B?
37 provider offerings are linked to the default version.
What models should I compare Llama 3.3 70B with?
Nearby ranked alternatives include Llama 3.3 Nemotron Super 49B, Qwen2.5 VL 72B, MiniCPM SALA.