NEWMEN
← back to all models

Meta: Llama 3.3 70B Instruct

meta-llama/llama-3.3-70b-instruct

Metaopen-weight131K context14 providersIntelligence 58.4

Cheapest provider

$0.10 / 1M

DeepInfra

Fastest provider (p95)

435 tok/s

Groq

Intelligence (composite)

58.4

MMLU-Pro · HumanEval · math · GPQA

Per-provider performance

Latency / throughput / uptime / price measured across providers over the last 30 minutes of live traffic. This is what proves “sourced cheapest” — Atlas mode draws on these per call to serve the cheapest path that holds quality.

ProviderQuantInput $/1MOutput $/1MLatency p50 / p95Throughput p50 / p95Uptime 30mSuccess
DeepInfraq8· fp8$0.1000$0.3200436ms / 2748ms16 / 30 tok/s98.72%98.7%
Inceptronq8· fp8$0.1200$0.3800543ms / 3056ms28 / 43 tok/s98.51%96.9%
AkashMLq8· fp8$0.1300$0.4000463ms / 1439ms36 / 69 tok/s99.49%99.5%
Nebiusq8· fp8$0.1300$0.4000367ms / 1737ms27 / 42 tok/s90.74%90.7%
Novitafull· bf16$0.1350$0.4000613ms / 1160ms32 / 50 tok/s99.71%90.0%
Parasailq8· int8$0.2200$0.5000679ms / 2107ms27 / 39 tok/s97.47%97.1%
Cloudflareq8· fp8$0.2930$2.2530513ms / 1723ms28 / 43 tok/s98.87%98.9%
SambaNovafull· bf16$0.4500$0.9000614ms / 2904ms60 / 260.4 tok/s96.70%98.4%
Groqundisclosed$0.5900$0.7900242ms / 666ms190 / 434.8 tok/s99.69%98.7%
SambaNovafull· bf16$0.6000$1.2000614ms / 2904ms60 / 260.4 tok/s98.41%98.4%
WandBfull· fp16$0.7100$0.7100314ms / 794ms47 / 89 tok/s100.00%100.0%
Googleundisclosed$0.7200$0.7200334ms / 1078ms22 / 99.1 tok/s90.04%90.0%
Googleundisclosed$0.7200$0.7200334ms / 1078ms22 / 99.1 tok/s100.00%90.0%
Togetherq8· fp8$1.0400$1.04001125ms / 5929ms22 / 81.8 tok/s99.30%99.3%

“—” means live telemetry hasn’t accumulated enough recent traffic for that endpoint. “undisclosed” means the provider serves the model but doesn’t expose the quantization label (typically running fp8 / int8 internally).

Intelligence breakdown

Composite score is a weighted average of public benchmarks (30% MMLU-Pro, 25% code pass@1, 25% math, 20% GPQA). Numbers come from model cards and the Artificial Analysis intelligence harness; missing components are renormalised over what’s present.

MMLU-Pro

68.9

broad reasoning

Code

33.3

pass@1 (HumanEval / LiveCodeBench)

MATH

77.0

math accuracy

GPQA Diamond

50.5

hard reasoning

Source: Meta Llama 3.x cards — Llama 3.3 70B (MMLU-Pro 68.9, GPQA-Diamond 50.5, LiveCodeBench 33.3); llama.com

How Atlas mode sources Meta: Llama 3.3 70B Instruct

  • Strict mode — pin Meta: Llama 3.3 70B Instruct exactly and we pass it straight through, sourced from the cheapest provider above. The same model, no substitutions — currently DeepInfra at $0.10/1M.
  • Atlas mode — the default. Each call is auto-optimized for the cheapest path that holds quality, at least 5% off going direct from call one and climbing as it ramps. You always see which model served the call and exactly what you saved — thumbs-down anything you don’t like for a full refund.