Meta
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. Muse Spark 1.1 is a multimodal reasoning model built for agentic and coding work, with a 1M self-managed context window, native primary-agent and subagent orchestration, and MCP support. It tops tool-use benchmarks (88.1 on MCP Atlas) though it trails the leaders on pure coding accuracy. Meta remains the most important player for teams that need on-premise deployments, fine-tuning, or zero marginal inference cost, and now also competes on low-cost hosted agentic inference.
Service status
Loading…
Founded
2004 (Meta); 2013 (FAIR)
Headquarters
Menlo Park, CA
CEO
Mark Zuckerberg
Models
3 active
Key Products
Strengths
- ✓Muse Spark 1.1 low-cost agentic API ($1.25/$4.25)
- ✓Fully open-source Llama family
- ✓Free to self-host and fine-tune
- ✓10M token context (Scout)
- ✓No per-token cost for open weights
Meta Models
| Model | Input / 1M | Output / 1M | Context | Capabilities |
|---|---|---|---|---|
| Muse Spark 1.1 | 1.25 | 4.25 | 1M | text, vision, tool-use, code, reasoning |
| Llama 4 Scout | Free | Free | 10M | text, vision, code |
| Llama 4 Maverick | Free | Free | 1M | text, vision, code |
Prices per 1M tokens in USD. See the full pricing guide for detailed analysis.
Benchmark Scores
| Model | SWE-bench | MMLU-Pro | HumanEval | GPQA Diamond | MATH |
|---|---|---|---|---|---|
| Llama 4 Scout | N/A | 85.9 | 88.4 | 56.2 | 79.8 |
| Llama 4 Maverick | 24.0 | 89.3 | 91.7 | 64.1 | 86.7 |