Skip to content
All systems operational0 AI providers monitored, polled every 2 minutes
Live status

Meta has positioned itself as the champion of open-source AI. Their Llama 4 family includes Scout (with a record 10 million token context window) and Maverick (which competes with proprietary mid-tier models on benchmarks), all free to download and self-host under the Llama Community License. On July 9, 2026 Meta shifted strategy with Muse Spark 1.1, its first paid API model, opened to US developers at $1.25 input / $4.25 output per 1M tokens, roughly a quarter of OpenAI and Anthropic list rates. On August 5, 2026 it followed with Muse Code, a terminal coding agent for macOS and Linux, and Muse Spark 1.2, a coding-focused update co-trained inside the Muse Code harness. Standard 1.2 pricing holds at $1.25/$4.25, but the new muse-spark-1.2-contributor tier serves the same weights at $0.10/$0.20 in exchange for training rights on your data, currently the cheapest frontier-adjacent coding path on the market. Muse Spark 1.3 followed on September 2, 2026 with every rate held from 1.2, Standard at $1.25/$4.25 with $0.15 cached input and Contributor at $0.10/$0.20, so the pitch moved from price to consumption: roughly 20 percent fewer tool calls and 25 percent fewer tokens on the same agentic work, a discount that reaches the invoice without touching the price list. Meta reports DeepSWE 1.1 at 75.4, SWEAtlas CodeBase QnA at 59.4, Terminal-Bench tied with GPT-5.6 Sol at 88.8, and MRCR at 98.5 and 98.1 across the long-context bands, with the caveat that the 1.3 column runs at max reasoning against 1.2 at xhigh. On October 12, 2026 Meta moved up to the frontier price tier with Muse Wonder, served on the Meta Model API as muse-wonder-1.0 at $4 / $20 per 1M tokens with $0.20 cached input and a 1M context window, the same rate card as Claude Opus 5.5; the stronger Muse Wonder Full stays inside Muse Code subscriptions, and it is the variant Artificial Analysis scores 56 on its Intelligence Index, tied for second. Meta had promised an open-weight Muse Spark through the 1.2 and 1.3 launches without a date, and that promise landed on October 15, 2026 as Muse Spark Open: a sparse mixture-of-experts with 816 billion total parameters and 42 billion active, natively multimodal, 1M context, published under the Muse Model License rather than Apache 2.0, with weights on Hugging Face, Amazon SageMaker AI, Google Cloud Model Garden, and Microsoft Foundry Models. Hosted inference runs through Baseten, Fireworks AI, Groq, Modal, Nebius, OpenRouter, and Together AI at $1.25 / $4.25 with $0.15 cached input, the same rate as the Muse Spark Standard tier, and Meta has not put it on its own Model API. Meta worked with Apple so it runs on two 512GB M5 Ultra Mac Studios, roughly a 1TB unified-memory floor, and calls it the most capable open-weight model from a US lab; the launch post publishes its benchmark section as a chart rather than a table. On November 9, 2026 Meta filled in the bottom of the lineup with Muse Ember, served as muse-ember-1.0 and described by Meta as its smallest, most efficient model, built for high-volume agentic workflow execution. It prices at $0.10 per 1M input tokens, $0.004 cached, and $0.50 output for prompts up to 200,000 input tokens, tripling to $0.30, $0.012, and $1.50 above that threshold, on a 1M context window with text, image, audio, and video input. The floor matches Claude Haiku 5.5 and GPT-6 Luna on list price while undercutting both on cached input, and Meta reports an Artificial Analysis Intelligence Index of 36 at max effort, level with Haiku 5.5 and ahead of Luna at 28, at roughly $0.13 per Index task. The gap it admits to is coding: FrontierCode 2.0-AA comes in at 27.2 against 34.9 for Haiku 5.5. Ember is on the Meta Model API and included in every Muse Code plan, and Meta has published no weights for it. Meta remains the most important player for teams that need on-premise deployments, fine-tuning, or zero marginal inference cost, and now also competes hard on low-cost hosted agentic inference.

Service status

Loading…

Latest news

Loading…

All news

Recent GitHub activity

Loading…

From /api/agents/opportunities, refreshed daily 13:30 UTC.

Founded

2004 (Meta); 2013 (FAIR)

Headquarters

Menlo Park, CA

CEO

Mark Zuckerberg

Models

11 active

Key Products

Muse WonderMuse Wonder Full (Muse Code)Muse EmberMuse Spark OpenMuse Spark 1.3Muse Spark 1.2Muse CodeMuse Spark 1.1Muse Glimmer 30BLlama 4 ScoutLlama 4 MaverickMeta AI Assistant

Strengths

  • ✓Muse Spark Open ships 816B open weights at 42B active under the Muse Model License
  • ✓Muse Ember at $0.10 / $0.50 up to a 200K prompt with $0.004 cached input
  • ✓Muse Spark 1.3 uses roughly 20 percent fewer tool calls and 25 percent fewer tokens than 1.2
  • ✓Contributor tier at $0.10/$0.20, cheapest frontier-adjacent coding path
  • ✓Muse Code terminal agent co-trained with the model
  • ✓Fully open-source Llama family
  • ✓10M token context (Scout)

Meta Models

ModelInput / 1MOutput / 1MContextCapabilities
Muse Ember0.100.501Mtext, vision, audio, video, tool-use, code, reasoning
Muse Wonder4.0020.001Mtext, tool-use, code, reasoning
Muse Spark Open1.254.251Mtext, vision, tool-use, code, reasoning
Muse Glimmer 30B0.351.50131Ktext, vision, tool-use, code, reasoning
Muse Spark 1.31.254.251Mtext, vision, tool-use, code, reasoning
Muse Spark 1.3 Contributor0.100.201Mtext, vision, tool-use, code, reasoning
Muse Spark 1.21.254.251Mtext, vision, tool-use, code, reasoning
Muse Spark 1.2 Contributor0.100.201Mtext, vision, tool-use, code, reasoning
Muse Spark 1.11.254.251Mtext, vision, tool-use, code, reasoning
Llama 4 ScoutFreeFree10Mtext, vision, code
Llama 4 MaverickFreeFree1Mtext, vision, code

Prices per 1M tokens in USD. See the full pricing guide for detailed analysis.

Benchmark Scores

ModelSWE-benchMMLU-ProHumanEvalGPQA DiamondMATHOSWorld 2.0BrowseCompFrontierCode v1.1Terminal-Bench 4.0Humanity's Last Exam (tools)
Muse WonderN/AN/AN/AN/AN/AN/AN/AN/A49.5N/A
Muse Spark OpenN/AN/AN/AN/AN/AN/AN/AN/A6.6N/A
Muse Glimmer 30B76.0N/AN/A83.5N/AN/AN/AN/AN/AN/A
Llama 4 ScoutN/A85.988.456.279.8N/AN/AN/AN/AN/A
Llama 4 Maverick24.089.391.764.186.7N/AN/AN/AN/AN/A
Muse Spark 1.3N/AN/AN/A94.1N/AN/AN/AN/A10.6N/A
Muse Spark 1.286.688.3N/AN/AN/AN/AN/AN/A6.1N/A
Muse Spark 1.182.088.7N/A91.2N/AN/AN/AN/AN/AN/A
Muse EmberN/AN/AN/AN/AN/AN/AN/AN/A39.4N/A

Comparisons

Other Providers

Visit MetaCompare ModelsAll Models