MMeta · Released Dec 6, 2024
Llama 3.3 70B Instruct
The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out). The Llama 3.3 instruction tuned text only model...
- Released
- Dec 6 2024
- Context window
- 131K 16K out
- Input / 1M tokens
- $0.10
- Output / 1M tokens
- $0.32
Where to run it
10 providers compared
Cheapest: DeepInfra at $0.10 / $0.32 per 1M input / output tokens — 85% less than the priciest option.
| Provider | Input | Output | Cache read | Context | Max output | Uptime | vs. best |
|---|---|---|---|---|---|---|---|
| DeepInfraFP8 | $0.10 | $0.32 | — | 131K | 16K | 100%24h | Cheapest |
| NovitaBF16 | $0.14 | $0.40 | — | 12K | 11K | 92.8%24h | +30% |
| AkashMLFP8 | $0.20 | $0.52 | $0.10 | 131K | 128K | 100%24h | +81% |
| ParasailFP8 · no tool calling | $0.22 | $0.50 | $0.11 | 131K | 16K | 98.4%24h | +87% |
| SambaNovano tool calling | $0.45 | $0.90 | — | 131K | 3K | 100%24h | +263% |
| Groq | $0.59 | $0.79 | $0.29 | 131K | 33K | 100%24h | +313% |
| CoreWeaveFP16 | $0.71 | $0.71 | $0.71 | 128K | 115K | 100%24h | +358% |
| Googleus-central1 | $0.72 | $0.72 | — | 128K | 8K | — | +365% |
| Googleno tool calling | $0.72 | $0.72 | — | 128K | 115K | — | +365% |
| CloudflareFP8 · no tool calling | $0.29 | $2.25 | — | 24K | 22K | 99.9%24h | +405% |
| Together | $1.04 | $1.04 | — | 131K | 2K | 92.1%24h | +571% |
Value
Price vs. performance
Llama 3.3 70B Instruct Other Meta Other labs Value frontier · GPQA Diamond vs. input price per 1M tokens (log scale) · Drag to pan, pinch or Ctrl + scroll to zoom, double-click to zoom in.
Benchmarks
7 benchmark scores
| Benchmark | Score | Rank | Leader |
|---|---|---|---|
| SciCodeSoftware engineering · Research-coding problems from the natural sciences | 26.0% | #98of 102 | Claude Opus 5.566.9% |
| Fiction.LiveBenchWriting & content · Understanding long creative writing (16k tokens) | 33.3% | #22of 24 | GPT-597.2% |
| GPQA DiamondReasoning & knowledge · PhD-level biology, chemistry and physics questions | 47.4% | #108of 122 | Gemini 4 Argon96.4% |
| SimpleBenchReasoning & knowledge · Common-sense trick questions about space, time and people | 19.9% | #52of 58 | Claude Fable 581.9% |
| CritPtReasoning & knowledge · Open-ended, research-level physics problems | 0.0% | #92of 111 | GPT-5.6 Sol32.3% |
| LMCAReasoning & knowledge · Judge the quality of conceptual arguments like experts | 17.5% | #98of 123 | Claude Sonnet 5.568.3% |
| OTIS Mock AIMEMaths · Competition-style problems harder than MATH Level 5 | 5.1% | #113of 120 | Gemini 4 Argon100.0% |
Specs
Details
- Model ID
meta-llama/llama-3.3-70b-instruct- Developer
- Meta
- Release date
- Dec 6, 2024 ·
- Context window
- 131,072 tokens
- Max output
- 16,384 tokens
- API price
- $0.10 / $0.32 per 1M input / output tokens
- Providers
- 10
- Input types
- text
- Reasoning
- No
- Tool calling
- Yes
- Open weights
- Yes
- Free tier
- No
Release date, prices and providers from OpenRouter's catalog; buying directly from a provider can cost a little differently. Uptime is the share of OpenRouter requests each provider served over the last day. Benchmark scores are the best result per benchmark from Epoch AI, licensed CC BY 4.0. Rank counts every model tracked here with a score.