MDL
Llama 3.3 Instruct 70B
Meta
Verified against Artificial Analysis · Jul 13, 2026
Capability
45
percentile index
Price /M
$0.58 / $0.71
in / out
Context
—
Avg score
32.4
11 benchmarks
AA Intelligence
9.4
index
Speed
93 tok/s
TTFT 0.6s
Released
Dec 6, 2024
About
Llama 3.3 Instruct 70B is a open-weights model in the Llama family from Meta. Benchmarked on 11 evals, averaging 32.4.
Benchmark scores · 11
avg 32.4 — higher is betterThe research behind this model
all papers mentioning it →Agentic Abstention: Do Agents Know When to Stop Instead of Act?▲ 146 on HF · Jun 27, 2026Puzzle: Distillation-Based NAS for Inference-Optimized LLMs2 cites · Nov 28, 2024Strategies for Guiding LLMs to Use Software Design Patterns: A Case of SingletonMay 26, 2026Understanding Conversational Patterns in Multi-agent Programming: A Case Study on Fibonacci Game DevelopmentMay 22, 2026A-ProS: Towards Reliable Autonomous Programming Through Multi-Model FeedbackMay 18, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Output speed93 tok/s
Latency (TTFT)0.6s
Price in / out$0.58 / $0.71 per 1M
Blended price$0.612 / 1M
Coding index11.9
Math index7.7
Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →
History
full ledger →Capability over time
Append-only ledger — every observed change to this model's numbers.
API providers
Artificial Analysis$0.71/M