Meta Llama 3 70B Instruct
Apr 2024
Meta Llama 3 70B InstructCompare inon
Meta Llama 3 70B Instruct measured on 3 of 104 Chat & Knowledge benchmarks3
BenchmarkByScoreRank
Legacy · saturated or no longer reported2
No benchmark matches
Alternatives?
ModelScoreCostPer taskWhat changes
CheaperGPT-6 Lunamax×2.2-38%$0.068Function calling only with none · Chat Completions+3BetterGPT-5.6 Lunalow×2+38%$0.01
OpenQwen3.6 27Bunstated×2.1+315%$0.60 / $3.60
CheaperGLM 5.3 FlashmaxOpen×2.2-27%$0.28Memory (4-bit): 42 GB → 193 GB+5
CheaperClaude Haiku 5.5max×2.2-38%$0.21Remove sampling parameters+5ModelScoreCostPer taskWhat changes
GPT-6 Lunamax×2.2-38%$0.068Function calling only with none · Chat Completions+3GLM 5.3 FlashmaxOpen×2.2-27%$0.28Memory (4-bit): 42 GB → 193 GB+5Claude Haiku 5.5max×2.2-38%$0.21Remove sampling parameters+5Qwen 3.5 Plusunstated×2.1-21%$0.12 / $0.69Qwen3.7 Flashunstated×2-83%$0.03 / $0.13ModelScoreCostPer taskWhat changes
GPT-5.6 Lunalow×2+38%$0.01GPT OSS 20BhighOpen+13%-89%$0.011
Gemini 3.1 Flash Litehigh×2+73%$0.06GPT-6 Lunamax×2.2-38%$0.068Function calling only with none · Chat Completions+3GPT OSS 120BhighOpen+87%-82%$0.092Memory (4-bit): 42 GB → 70 GBModelScoreCostPer taskWhat changes
Qwen3.6 27BunstatedOpen×2.1+315%$0.60 / $3.60Qwen3.6 35B A3BnoneOpen×2.1+71%$0.25 / $1.49Qwen 3.5 35B A3BunstatedOpen×2.1+112%$0.25 / $2Qwen 3.5 9BunstatedOpen+95%-72%$0.08 / $0.13Gemma 4 31BminimalOpen+87%-51%$0.10 / $0.34ModelScoreCostPer taskWhat changes
Muse Sparkunstated×2.2—$1.25 / $4.25Only n=1 · Meta / Chat Completions+1Muse Spark Open 816B A42B×2.2—$1.25 / $4.25Context: not verifiedLlama 4 Maverick 17B 128E Instruct FP8Open+65%—$0.20 / $0.80Memory (4-bit): 42 GB → 241 GBLlama 4 ScoutunstatedOpen+28%—$0.10 / $0.30Memory (4-bit): 42 GB → 65 GBLlama 3.1 405B Instruct+26%——Llama 3.3 70B InstructunstatedOpen+17%—$0.10 / $0.32Llama 3.1 70B InstructunstatedOpen+9%—$0.40 / $0.40Llama 3.2 90B Vision InstructOpensame—$0.35 / $0.40Memory (4-bit): 42 GB → 53 GBAll Meta models →Run it yourself
estimateFits on✗ 8 GB (phone, laptop)✗ 16 GB✗ 24 GB GPU✓ 48 GB GPU · compressed✓ 80 GB GPU · 8-bit
LicenceLlama 3 licencecommercial use with conditions
Benchmarks
in Chat & KnowledgeGPQA Diamond · Epoch40.6#114
MATH Level 522.6#29
OTIS Mock AIME 2024 20254.3#115
FAQ
How much does Meta Llama 3 70B Instruct cost?›
$0.30 per 1M input tokens and $0.40 per 1M output tokens (cheapest host: DeepInfra).
Is Meta Llama 3 70B Instruct free to use?›
There is no free API; the cheapest paid price is $0.30 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Meta Llama 3 70B Instruct, and what is it best at?›
#114 of 133 in Chat & Knowledge.
Is Meta Llama 3 70B Instruct better than Qwen 2 72B Instruct?›
On GPQA Diamond · Epoch, Meta Llama 3 70B Instruct scores 40.6 vs 40.8 — about the same.
How can I use Meta Llama 3 70B Instruct?›
Through the API of DeepInfra.
What is the context window of Meta Llama 3 70B Instruct?›
8K tokens.
When was Meta Llama 3 70B Instruct released? Is there a newer version?›
Released 2024-04-17. It is the current version in its line.
Is Meta Llama 3 70B Instruct open source?›
Yes, the weights are open (llama3 licence).
Can I run Meta Llama 3 70B Instruct locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Meta Llama 3 70B Instruct read images?›
No, it accepts text only.
What is Meta Llama 3 70B Instruct?›
Meta Llama 3 70B Instruct is an open-weights language model from Meta, released Apr 2024.
What are the best alternatives to Meta Llama 3 70B Instruct?›
GPT-6 Luna (cheaper), GPT-5.6 Luna (better), Qwen3.6 27B (open weights).