Skip to main content

Llama 4 Scout 17B 16E Instruct

Llama 4 Scout 17B Instruct (16E) is a mixture-of-experts (MoE) language model developed by Meta, activating 17 billion parameters out of a total of 109B. It supports native multimodal input (text and image) and multilingual output (text and code) across 12 supported languages. Designed for assistant-style interaction and visual reasoning, Scout uses 16 experts per forward pass and features a context length of 10 million tokens, with a training corpus of ~40 trillion tokens. Built for high efficiency and local or commercial deployment, Llama 4 Scout incorporates early fusion for seamless modality integration. It is instruction-tuned for use in multilingual chat, captioning, and image understanding tasks. Released under the Llama 4 Community License, it was last trained on data up to August 2024 and launched publicly on April 5, 2025.

Input / output price
$0.10 / $0.30 per 1M
Context
328K tokens
Released
Apr 5, 2025
Agent setup

Paste into Claude Code, Cursor, or Codex

Help me set up the Respan gateway: fetch and follow https://www.respan.ai/docs/agent-setup.md. Use model "llama-4-scout-17b-16e-instruct".

Read docs

Providers

3 providers, cheapest first, priced per 1M tokens after gateway discounts.

ProviderInputOutput
DeepInfradeepinfra/llama-4-scout-17b-16e-instruct
$0.10/M$0.30/M
Bedrockbedrock/llama-4-scout-17b-16e-instruct
$0.17/M$0.66/M
Novitanovita/llama-4-scout-17b-16e-instruct
$0.18/M$0.59/M

Is a provider down? AWS Bedrock status

Pricing

Averaged across 3 providers, in USD per 1M tokens after gateway discounts.

Average input
$0.15 per 1M
Average output
$0.52 per 1M

Cost by volume

Total cost from 0 to 1T tokens on the cheapest and most expensive provider.

Highest$0.18/MNovitaLowest$0.10/MDeepInfra

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 better

44tokens/sacross providers

Faster than 60% of the other 599 models on the gateway

Time to first token

Lower is better

1.0sacross providers

Faster than 71% of the other 152 models on the gateway

Response latency

Lower is better

0.9sacross providers

Faster than 71% of the other 780 models on the gateway

Benchmarks

The Respan Index plus 8 published scores from 4 sources, each linked to where it was published.

11.2/ 100

Respan Index#168 of 176

176 ranked models. The shaded band is this model's 95% interval, 9.2 to 15.5. DeepSeek V3 and GPT-4.1 nano fall inside it, so the gap is within the margin of error.
Llama 4 Scout 17B 16E Instruct by category: Coding 0.7 (estimated), Agents 3.7 (estimated), Reasoning 16.8, Knowledge & language 6.9, Human preference 34.6
  1. 168Llama 4 Scout11.2
  2. 167DeepSeek V313.2
  3. 169GPT-4.1 nano11.1

Published scores

BenchmarkScore
GPQA Diamondhuggingface.co57.2%
GPQA DiamondEpoch AI51.8%

Compare Llama 4 Scout 17B 16E Instruct: Llama 4 Scout vs Muse Wonder Full, Llama 4 Scout vs Muse Wonder, Llama 4 Scout vs Muse Ember, Llama 4 Scout vs Muse Spark 1.3, Llama 4 Scout vs Muse Spark 1.1, Llama 4 Scout vs Muse Spark 1.2

Apps

Use Llama 4 Scout 17B 16E 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 4 Scout 17B 16E Instruct FAQ

Short answers, built from the data on this page.

Llama 4 Scout 17B 16E Instruct is a model from Meta, released on Apr 5, 2025. Llama 4 Scout 17B Instruct (16E) is a mixture-of-experts (MoE) language model developed by Meta, activating 17 billion parameters out of a total of 109B. It supports native multimodal input (text and image) and multilingual output (text and code) across 12 supported languages. Designed for assistant-style interaction and visual reasoning, Scout uses 16 experts per forward pass and features a context length of 10 million tokens, with a training corpus of ~40 trillion tokens. Built for high efficiency and local or commercial deployment, Llama 4 Scout incorporates early fusion for seamless modality integration. It is instruction-tuned for use in multilingual chat, captioning, and image understanding tasks. Released under the Llama 4 Community License, it was last trained on data up to August 2024 and launched publicly on April 5, 2025.

On the Respan gateway, Llama 4 Scout 17B 16E Instruct costs from $0.10 per 1M input tokens and $0.30 per 1M output tokens on DeepInfra, after gateway discounts. Averaged across its 3 providers, it costs $0.15 per 1M input tokens and $0.52 per 1M output tokens.

Llama 4 Scout 17B 16E Instruct is available through 3 providers: DeepInfra, Bedrock, and Novita.

Llama 4 Scout 17B 16E Instruct accepts up to 328K 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-4-scout-17b-16e-instruct". To set it up with a coding agent, copy the agent setup prompt at the top of this page.

Llama 4 Scout 17B 16E Instruct scores 11.2 out of 100 on the Respan Index v1.1, ranking #168 of 176 models. By category: Coding 0.7 (estimated), Agents 3.7 (estimated), Reasoning 16.8, Knowledge & language 6.9, and Human preference 34.6. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by Respan.

Llama 4 Scout 17B 16E Instruct has 8 published benchmark scores, including GPQA Diamond 57.2%, MATH Level 5 62.3%, and MMLU-Pro 74.3%. 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.

Works with your entire stack

Use Respan with your favorite frameworks and tools.

AnthropicAI21 LabsxAIVercel AI SDK
LangChainOpenRouterAssemblyAICursor
LiteLLMGroqBasetenMem0
CohereCogneeLlamaIndexFireworks
DeepSeekLinkupHaystackTogether AI
PerplexityInference.netPostHogRubyLLM
SlackMastraAzure OpenAIMistral
AWS BedrockMoonshotHyperspellBAML
ModaInstructorGoogle GeminiParasail
ReductoZapierAgnoNebius
OpenAINovita AIReplicateSuperagent

Built for AI agents.
Break less.
Ship more.