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 is optimized for multilingual dialogue use cases and outperforms many of the available open source and closed chat models on common industry benchmarks. Supported languages: English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai. Model Card
- Input / output price
- $0.10 / $0.32 per 1M
- Context
- 131K tokens
- Released
- Dec 6, 2024
Paste into Claude Code, Cursor, or Codex
Help me set up the Respan gateway: fetch and follow https:/
Providers
8 providers, cheapest first, OpenRouter last, priced per 1M tokens after gateway discounts.
| Provider | Input | Output |
|---|---|---|
Concentrate concentrate/llama-3.3-70b-instruct | $0.10/M | $0.32/M |
Inference.net inference_net/llama-3.3-70b-instruct | $0.13/M | $0.40/M |
NovitaRetiring novita/llama-3.3-70b-instruct | $0.135/M | $0.40/M |
Parasail parasail/llama-3.3-70b-instruct | $0.22/M | $0.50/M |
IO.net ionet/llama-3.3-70b-instruct | $0.4786/M | $1.0946/M |
Azure Foundry azure_foundry/llama-3.3-70b-instruct | $0.71/M | $0.71/M |
Bedrock bedrock/llama-3.3-70b-instruct | $0.72/M | $0.72/M |
OpenRouter openrouter/llama-3.3-70b-instruct | $0.71/M | $0.71/M |
Is a provider down? AWS Bedrock status, OpenRouter status
Pricing
Averaged across 8 providers, in USD per 1M tokens after gateway discounts.
- Average input
- $0.40 per 1M
- Average output
- $0.61 per 1M
- Average cache read
- $0.35 per 1M
Cost by volume
Total cost from 0 to 1T tokens on the cheapest and most expensive provider.
Throughput and latency
30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.
Output speed
Higher is better4.2tokens/sacross providers
Response latency
Lower is better0.5sacross providers
Benchmarks
The Respan Index plus 9 published scores from 6 sources, each linked to where it was published.
10.9/ 100
Respan Index#170 of 176
- 170Llama 3.3 70B Instruct10.9
- 169GPT-4.1 nano11.1
- 171GPT-4o mini10.7
Published scores
| Benchmark | Score | Source | Published |
|---|---|---|---|
| WeirdMLHåvard Tveit Ihle | 14.4% | Håvard Tveit Ihle | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| GPQA Diamondhuggingface.co | 50.5% | huggingface.co | Apr 5, 2025 |
| GPQA DiamondEpoch AI | 47.4% | Epoch AI | Oct 5, 2026 |
| SimpleBenchSimpleBench | 19.9% | SimpleBench | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| MATH Level 5Epoch AI | 41.6% | Epoch AI | Oct 5, 2026 |
| OTIS Mock AIME 2024-2025Epoch AI | 5.1% | Epoch AI | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| MMLU-Prohuggingface.co | 68.9% | huggingface.co | Apr 5, 2025 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| LMArena EloLMArena | 1318 Elo · #166 of 191 | LMArena | Nov 19, 2026 |
Compare Llama 3.3 70B Instruct: Llama 3.3 70B Instruct vs Llama 3.1 70B Instruct, Llama 3.3 70B Instruct vs Muse Wonder Full, Llama 3.3 70B Instruct vs Muse Wonder, Llama 3.3 70B Instruct vs Muse Ember, Llama 3.3 70B Instruct vs Muse Spark 1.3, Llama 3.3 70B Instruct vs Muse Spark 1.1
Apps
Use Llama 3.3 70B Instruct in the tools you already run. Each guide points the app at the Respan gateway.
Setup guides for more apps are in the integration docs.
Llama 3.3 70B Instruct FAQ
Short answers, built from the data on this page.
Llama 3.3 70B Instruct is a model from Meta, released on Dec 6, 2024. 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 is optimized for multilingual dialogue use cases and outperforms many of the available open source and closed chat models on common industry benchmarks. Supported languages: English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai. Model Card.
On the Respan gateway, Llama 3.3 70B Instruct costs from $0.10 per 1M input tokens and $0.32 per 1M output tokens on Concentrate, after gateway discounts. Averaged across its 8 providers, it costs $0.40 per 1M input tokens and $0.61 per 1M output tokens.
Llama 3.3 70B Instruct is available through 8 providers: Concentrate, Inference.net, Novita, Parasail, IO.net, Azure Foundry, and 2 more.
Llama 3.3 70B Instruct accepts up to 131K tokens of context on its largest provider route.
Point any OpenAI-compatible SDK at https://gateway.respan.ai/v1 with your Respan API key and set the model to "llama-3.3-70b-instruct". To set it up with a coding agent, copy the agent setup prompt at the top of this page.
Llama 3.3 70B Instruct scores 10.9 out of 100 on the Respan Index v1.1, ranking #170 of 176 models. By category: Coding 0.7, Agents 4.4, Reasoning 16.4, Knowledge & language 4.3, and Human preference 34.0. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by Respan.
Llama 3.3 70B Instruct has 9 published benchmark scores, including WeirdML 14.4%, BALROG 23%, and GPQA Diamond 50.5%. Each score links to where it was published; the data is collected by Respan.
Certified where it counts.
Built to meet the security and privacy standards that enterprise and healthcare teams require.
ISO 27001
The internationally recognized standard for information security management.
SOC 2
Secure, compliant management of your data across all of our systems.
GDPR
Operated under GDPR, the world's strictest standard for data privacy.
HIPAA
HIPAA compliant, with a BAA available for healthcare teams.