Meta

Model lab · 16 models · also sold by 19 other providers · meta.com

Meta models

			In / out per 1M			Cheapest route
Muse Spark Open 1M Muse Spark Open is a coding-optimized open-weight model.	$1.25	$4.25	$1.25 / $4.25	$2	1M	on Modal	·	Muse Spark Open is a coding-optimized open-weight model.
Muse Wonder 1M Muse Wonder is a frontier reasoning model from Meta, designed for complex personal and professional agentic tasks.	$4	$20	$4 / $20	$8	1M	Lab list price	·	Muse Wonder is a frontier reasoning model from Meta, designed for complex personal and professional agentic tasks.
Muse Spark 1.3 1M Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows.	$1.25	$4.25	$1.25 / $4.25	$2	1M	Lab list price	·	Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows.
Muse Spark 1.3 Contributor 1M Muse Spark 1.3 Contributor is the cost-efficient contributor tier of Meta’s multimodal reasoning model for experimentation, learning, and early-stage agentic, multi-agent, and coding workflows.	$0.1	$0.2	$0.1 / $0.2	$0.125	1M	Lab list price	·	Muse Spark 1.3 Contributor is the cost-efficient contributor tier of Meta’s multimodal reasoning model for experimentation, learning, and early-stage agentic, multi-agent, and coding workflows.
Muse Spark 1.2 Contributor 1M Muse Spark 1.2 contributor tier is a reasoning model from Meta designed for developers who want to start building at an even lower cost.	$0.1	$0.2	$0.1 / $0.2	$0.125	1M	Lab list price	·	Muse Spark 1.2 contributor tier is a reasoning model from Meta designed for developers who want to start building at an even lower cost.
Muse Glimmer 30B 131K Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware.	$0.3	$1.2	$0.3 / $1.2	$0.525	131K	on DeepInfra (bf16)	·	Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware.
Muse Spark 1.2 1M Muse Spark 1.2 is a reasoning model from Meta, designed for complex agentic tasks.	$1.25	$4.25	$1.25 / $4.25	$2	1M	Lab list price	·	Muse Spark 1.2 is a reasoning model from Meta, designed for complex agentic tasks.
Muse Spark 1.1 1M Muse Spark 1.1 is a multimodal reasoning model from Meta, built for agentic tasks.	$1.25	$4.25	$1.25 / $4.25	$2	1M	Lab list price	·	Muse Spark 1.1 is a multimodal reasoning model from Meta, built for agentic tasks.
Llama Guard 4 12B 164K Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification.	$0.144	$0.144	$0.144 / $0.144	$0.144	164K	on DeepInfra (bf16-flex)	·	Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification.
Llama 4 Maverick 524K Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...	$0.188	$0.653	$0.188 / $0.653	$0.304	524K	on DigitalOcean	·	Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...
Llama 4 Scout 1.31M 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.	$0.08	$0.24	$0.08 / $0.24	$0.12	1.31M	on DeepInfra (fp8-flex)	·	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.
Llama 3.3 70B Instruct 131K The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).	$0.1	$0.32	$0.1 / $0.32	$0.155	131K	on DeepInfra (turbo)	·	The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).
Llama 3.2 1B Instruct 60K Llama 3.2 1B is a 1-billion-parameter language model focused on efficiently performing natural language tasks, such as summarization, dialogue, and multilingual text analysis.	$0.027	$0.201	$0.027 / $0.201	$0.071	60K	on Cloudflare Workers AI	·	Llama 3.2 1B is a 1-billion-parameter language model focused on efficiently performing natural language tasks, such as summarization, dialogue, and multilingual text analysis.
Llama 3.2 3B Instruct 131K Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization.	$0.05	$0.33	$0.05 / $0.33	$0.12	131K	on Parasail (bf16)	·	Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization.
Llama 3.1 70B Instruct 131K Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.	$0.72	$0.72	$0.72 / $0.72	$0.72	131K	on AWS Bedrock	·	Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.
Llama 3.1 8B Instruct 131K Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.	$0.02	$0.04	$0.02 / $0.04	$0.025	131K	on DeepInfra (fp8)	·	Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.

On other routes

Meta

Meta models

On other routes

About Meta

Meta on the price map

Model	In / out per 1M	Vs headline
Llama 3.1 8B Instruct fp8 · Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.	$0.02 / $0.04Same vs headline	Same	fp8 · Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.
Llama 3.1 8B Instruct Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.	$0.05 / $0.08+130% vs headline	+130%	Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.
Llama 3.2 1B Instruct Llama 3.2 1B is a 1-billion-parameter language model focused on efficiently performing natural language tasks, such as summarization, dialogue, and multilingual text analysis.	$0.027 / $0.201Same vs headline	Same	Llama 3.2 1B is a 1-billion-parameter language model focused on efficiently performing natural language tasks, such as summarization, dialogue, and multilingual text analysis.
Llama 4 Scout fp8-flex, fp8 · 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.	$0.08 / $0.24Same vs headline	Same	fp8-flex, fp8 · 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.
Llama 3.2 3B Instruct bf16 · Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization.	$0.05 / $0.33Same vs headline	Same	bf16 · Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization.
Llama 3.2 3B Instruct Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization.	$0.051 / $0.335+2% vs headline	+2%	Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization.
Llama Guard 4 12B bf16-flex, bf16 · Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification.	$0.144 / $0.144Same vs headline	Same	bf16-flex, bf16 · Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification.
Llama 4 Scout fp8 · 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.	$0.1 / $0.3+25% vs headline	+25%	fp8 · 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.
Llama 3.3 70B Instruct turbo, fp8 · The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).	$0.1 / $0.32Same vs headline	Same	turbo, fp8 · The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).
Llama Guard 4 12B bf16 · Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification.	$0.18 / $0.18+25% vs headline	+25%	bf16 · Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification.
Llama 3.1 8B Instruct fp8 · Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.	$0.152 / $0.287+643% vs headline	+643%	fp8 · Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.
Llama 3.1 8B Instruct bf16 · Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.	$0.22 / $0.22+780% vs headline	+780%	bf16 · Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors.
Llama 4 Scout fp8-priority, fp8 · 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.	$0.15 / $0.45+87% vs headline	+87%	fp8-priority, fp8 · 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.
Llama Guard 4 12B bf16-priority, bf16 · Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification.	$0.27 / $0.27+88% vs headline	+88%	bf16-priority, bf16 · Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification.
Llama 3.3 70B Instruct fp8 · The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).	$0.2 / $0.52+81% vs headline	+81%	fp8 · The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).
Llama 3.3 70B Instruct fp8 · The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).	$0.22 / $0.5+87% vs headline	+87%	fp8 · The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).
Llama 4 Maverick Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...	$0.188 / $0.653Same vs headline	Same	Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...
Llama 4 Scout us-east5 · 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.	$0.25 / $0.7+202% vs headline	+202%	us-east5 · 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.
Llama 4 Maverick fp8 · Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...	$0.35 / $1+69% vs headline	+69%	fp8 · Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...
Muse Glimmer 30B bf16 · Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware.	$0.3 / $1.2Same vs headline	Same	bf16 · Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware.