Skip to content

Meta · Chat

Llama 3.3 70B Instruct

Meta 🇺🇸ChatLlama familyOpen weights. One of its sources says the weights are publicly available.The maker currently plans to deprecate it within the next 180 days (31 Jan 2027). Plans can change.

This version of Llama-3.3-70b-instruct is also the FP8 quantized version of the original FP16 weights.

llminfo IQ
108.1
Market median price
$0.71 blended / 1M
Max context
131K
Listings
119

Evidence and sources

🤗 meta-llama/Llama-3.3-70B-Instruct70B paramsllminfo IQ 108.1, ranked among chat models. From 278 benchmarks. Plenty of data behind it.How llminfo IQ is computedCoding IQ 77.1, 95% interval 68–86. From 7 benchmark families across 7 sources. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.How Coding IQ is computedPoor value (31.6). Costs more per task than its benchmark results predict. From 7 benchmarks with measured task costs. A fair amount of data behind it.How Value for money worksReleased 26 Nov 2024 Training-data cutoff, as its sources state it Maker HQ: US (according to Meta's own website)
All rankings: #75 on NanoGPT recently #138 on OpenRouter today #155 on OpenRouter this month #157 on OpenRouter this weekSources: LMArena data, IBM watsonx.ai Foundation Models data docs, flex.ai API data, LiteLLM data, Models.dev data, OpenRouter data docs, ZenMux data, Hugging FacePrimary data from LMArena from IBM watsonx.ai Foundation Models: parameter count, description from flex.ai API: hugging face id from LiteLLM: deprecation date from Models.dev: weights availability from OpenRouter: knowledge cutoff from ZenMux: primary mode from Hugging Face: release date
Tags from its sources:multilingualinstructtoolsopen-sourceinstruction-tunedfunction-calling

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ

Advantages

Open weights available

The weights are public. 2 of the 7 peers that say either way have open weights.

Trade-offs

Deprecation scheduled

The maker plans to deprecate it on 31 Jan 2027, within the next 180 days.

We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 06:07 UTC.

Input & output

as stated by the model's providers
Text
Text

Capabilities

Supported operations and provider-stated features

ReasoningFunction callingStructured outputPrompt cachingWeb searchBatch API

Listings

 Flowdotmeta-llama/llama-3.3-70b-instruct:free · freeSource: FlowDot data66K$0.00$0.00
 OpenRoutermeta-llama/llama-3.3-70b-instruct:free · freeSource: Helicone Costs dataNot published by the provider$0.00$0.00
 NanoGPTmeta-llama/llama-3.3-70b-instructSources: Models.dev data, NanoGPT131K$0.05cache read $0.025$0.23
 Obscurifymeta-llama/llama-3.3-70b-instructSource: Models.dev data128K$0.0553$0.2542
 Eden AIInference by DeepInfradeepinfra/meta-llama/Llama-3.3-70B-InstructSource: Models.dev data131K$0.10$0.32
 Hugging FaceInference by DeepInframeta-llama/Llama-3.3-70B-InstructSource: Hugging Face Inference Providers data131K$0.10$0.32
 Kilometa-llama/llama-3.3-70b-instructSources: Kilo AI Gateway data, Models.dev data128K$0.10$0.32
 Meganovameta-llama/Llama-3.3-70B-InstructSources: MegaNova data, Models.dev data131K$0.10$0.30
 Lambda AIlambda_ai/llama3.3-70b-instruct-fp8 · fp8Source: LiteLLM data131K$0.12$0.30
 LlmgatewayInference by Nebius AI Studiometa-llama/Llama-3.3-70B-InstructSource: LLMGateway dataNot published by the provider$0.13$0.40

16 listings without a price are hidden.

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Multilingual

EKI EstonianSource: Keelemudelite mõõdupuu data docs
IndicParam42 related benchmarks · 42 results
Show 36 more resultsHide 36 results
PashtoBench10 related benchmarks · 10 results
Show 4 more resultsHide 4 results

Agentic

BFCL v2Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Gaia2Source: Gaia2 (Meta ARE) data · Sep 2025
Source: OpenRouter benchmark exports
llama-3.3-70b-instruct · 90-day results
Across 7 providers lowest 25.3% median 34.2% highest 40%OpenRouter's automatic routing 33%
Show results from 7 providers and OpenRouter's automatic routingHide results from 7 providers and OpenRouter's automatic routing
  • Nebius
  • Novita
  • AkashML
  • CoreWeave
  • OpenRouter's automatic routing
  • PrimeIntellect
  • Together
  • DeepInfra

Human preference

Arena ELO27 related benchmarks · 27 results

Math

MATHSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
MGSMSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
U-MATH7 related benchmarks · 21 results
algebra accuracy
Source: U-MATH and mu-MATH
Task set · algebra · gpt-4o-2024-08-06
differential_calc accuracy
Source: U-MATH and mu-MATH
Task set · differential_calc · gpt-4o-2024-08-06
integral_calc accuracy
Source: U-MATH and mu-MATH
Task set · integral_calc · gpt-4o-2024-08-06
multivariable_calculus accuracy
Source: U-MATH and mu-MATH
Task set · multivariable_calculus · gpt-4o-2024-08-06
overall accuracy
Source: U-MATH and mu-MATH
Task set · u_math · gpt-4o-2024-08-06
precalculus_review accuracy
Source: U-MATH and mu-MATH
Task set · precalculus_review · gpt-4o-2024-08-06
Show 3 more resultsHide 3 results
sequences_series accuracy
Source: U-MATH and mu-MATH
Task set · sequences_series · gpt-4o-2024-08-06

Coding

HumanEvalSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
SWE-rebench 2025-019 related benchmarks · 13 results
SWE-Effi Resolved Rate
Source: SWE-Effi data docs
Show 3 more resultsHide 3 results

Science

GPQASource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
GPQA Diamond14 results
Source: Epoch AI Benchmarking Hub data
Setting not stated
Source: OpenRouter benchmark exports
llama-3.3-70b-instruct · 90-day results
Across 12 providers lowest 45.6% median 47.8% highest 49.5%OpenRouter's automatic routing 48.2%
Show results from 12 providers and OpenRouter's automatic routingHide results from 12 providers and OpenRouter's automatic routing
  • Groq
  • Cloudflare
  • CoreWeave
  • AkashML
  • OpenRouter's automatic routing
  • SambaNova
  • Parasail
  • Together
  • Novita
  • PrimeIntellect
  • Crusoe
  • DeepInfra
  • Nebius

General

Factuality

LLM-AggreFact12 related benchmarks · 12 results
CNN
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, CNN, renderer-v1
all-11 mean
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, all-11 mean, renderer-v1
ClaimVerify
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, ClaimVerify, renderer-v1
ExpertQA
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, ExpertQA, renderer-v1
FactCheck
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, FactCheck, renderer-v1
LFQA
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, LFQA, renderer-v1
Show 6 more resultsHide 6 results
MediaS
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, MediaS, renderer-v1
MeetB
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, MeetB, renderer-v1
RAGTruth
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, RAGTruth, renderer-v1
REVEAL
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, REVEAL, renderer-v1
WiCE
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, WiCE, renderer-v1
XSum
Source: LLM-AggreFact data
Llama-3.3-70B-Instruct, grounded claim checker, XSum, renderer-v1

Safety

Vision

Long context

Instruction following

IFEvalSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim

Knowledge

Reasoning

Other

MedHELM33 related benchmarks · 33 results
Nejumi 428 related benchmarks · 28 results
ALT_バイアス
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
ALT_倫理・道徳
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
ALT_制御性
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
ALT_堅牢性
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
ALT_毒性
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
ALT_真実性
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
Show 22 more resultsHide 22 results
AVG_jaster_0shot
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
AVG_jaster_2shots
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_アプリケーション開発
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_コーディング
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_一般的知識
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_基礎的言語性能
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_専門的知識
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_応用的言語性能
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_情報検索
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_意味解析
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_抽象的推論
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_推論能力
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_数学的推論
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_構文解析
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_知識・質問応答
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_翻訳
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_表現
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_論理的推論
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
GLP_関数呼び出し
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
TOTAL_SCORE
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
アラインメント(ALT)_AVG
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
汎用的言語性能(GLP)_AVG
Source: Nejumi Leaderboard 4 data docs
meta-llama/Llama-3.3-70B-Instruct
Sayf16 related benchmarks · 16 results
Show 10 more resultsHide 10 results

Data credits: SEA-HELM leaderboard, AI Singapore (AI Products team), National University of Singapore. Published aggregate scores; results licence not stated (SEA-HELM code is MIT). · LMArena, CC BY 4.0 (licence) · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · Data by LLM Stats · EQ-Bench (Samuel J. Paech) — cite paech2023eqbench · DABstep (Adyen), CC BY 4.0 (licence) · Keelemudelite mõõdupuu: Independent LLM leaderboard for Estonian. Eesti Keele Instituut. https://moodupuu.eki.ee, CC BY 4.0. Latest per-benchmark results selected with partial-run flags retained. (licence) · FaithJudge, Vectara; Tamber et al., EMNLP 2025 Industry. Public numeric README leaderboard; results licence not separately established. · Galileo Agent Leaderboard, Apache-2.0 (licence) · OpenRouter public benchmark leaderboard and search-lane exports, CC BY 4.0. Fractional source scores converted to percentages; immutable dated snapshots. (licence) · Maheshwari et al., IndicParam, arXiv:2512.00333v3. Cited numeric scores under the owner's 2026-10-10 task-only restriction scope decision; no results licence established. Tasks, prompts and generations excluded. · LLM-AggreFact, Tang et al. Public numeric grounded checker balanced accuracy; results licence not separately established. · Stanford HELM (Stanford CRFM); scenario scores from Capabilities, Safety, Long Context, MedHELM, AIR-Bench, MMLU and Finance · Nejumi Leaderboard 4, Weights & Biases Japan; public publisher-selected numeric evaluation results · Open Arabic LLM Leaderboard (OALL), Apache-2.0 (licence) · PashtoBench v1.0, Jawad Noori. Public results/scores numbers; code MIT and data/ CC BY 4.0 grants do not name results/, so no results licence is asserted. · Phare (Giskard), CC BY 4.0 (licence) · Sayf-eval. Copyright (c) 2026 Qatar Computing Research Institute (QCRI). MIT License. (licence) · Shaberi v2.1, Shisa.AI, Apache-2.0. GPT-5.1-judge scores only; judges are not comparable. (licence) · SnakeBench (Greg Kamradt / ARC Prize Foundation), MIT (licence) · Swallow LLM Leaderboard v2, Swallow LLM team (Institute of Science Tokyo). Contents CC BY 4.0; Japanese post-trained task scores on the source's 0..1 scale. (licence) · SWE-Effi (Centre for Software Excellence), Apache-2.0; resolved rate converted to percent; resource figures are per-task means (licence) · Toloka, U-MATH and mu-MATH public leaderboard. MIT Space; generated solution outputs retain underlying model terms and are not ingested. (licence) · UzLiB, Tahrirchi (Abror Shopulatov). Public LEADERBOARD.md numbers; repository MIT covers software and documentation, no separate results licence stated and none asserted.

Also known as: azure_ai/llama-3.3-70b-instruct, cerebras-llama-3.3-70b, cerebras/llama-3.3-70b, crusoe/meta-llama/llama-3.3-70b-instruct, databricks/databricks-meta-llama-3-3-70b-instruct, deepinfra/meta-llama/llama-3.3-70b-instruct, fireworks_ai/accounts/fireworks/models/llama-v3p3-70b-instruct, gmi/meta-llama/llama-3.3-70b-instruct, gradient_ai/llama3.3-70b-instruct, hyperbolic/meta-llama/llama-3.3-70b-instruct, ionos/meta-llama/llama-3.3-70b-instruct, lambda_ai/llama3.3-70b-instruct-fp8, llama-3-3-70b, llama-3-3-70b-instruct, llama-3.3-70b, llama-3.3-70b-instruct-fp8, llama-3.3-70b:web, llama3-3-70b, llama3.3-70b, llama3.3:70b, meta-llama-3-3-70b-instruct, meta-llama-3.3-70b-instruct, meta-llama-3_3-70b-instruct, meta-llama/llama-3-3-70b-instruct, meta-llama/llama-3.3-70b-instruct, meta-llama/llama-3.3-70b-instruct:free, meta-llama/meta-llama-3.3-70b-instruct, meta.llama-3.3-70b-instruct, meta.llama3-3-70b-instruct-v1:0, meta/llama-3-3-70b, meta/llama-3.3-70b, meta/llama-3.3-70b-instruct, meta/llama-3.3-70b-instruct-maas, meta_llama/llama-3.3-70b-instruct, nebius/meta-llama/llama-3.3-70b-instruct, nim/meta/llama-3.3-70b-instruct, novita/llama-3.3-70b-instruct, novita/meta-llama/llama-3.3-70b-instruct, nscale/meta-llama/llama-3.3-70b-instruct, oci/meta.llama-3.3-70b-instruct, ovhcloud/meta-llama-3_3-70b-instruct, sambanova/meta-llama-3.3-70b-instruct, scaleway/llama-3.3-70b-instruct, scaleway/meta/llama-3.3-70b-instruct, snowflake-llama3.3-70b, snowflake/llama3.3-70b, snowflake/snowflake-llama-3.3-70b, tee/llama3-3-70b, us.meta.llama3-3-70b-instruct-v1:0, vercel_ai_gateway/meta/llama-3.3-70b, vertex_ai/meta/llama-3.3-70b-instruct-maas, wandb/meta-llama/llama-3.3-70b-instruct, watsonx/meta-llama/llama-3-3-70b-instruct

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works