Muse Glimmer 30B is Meta's cheapest model at $0.35/1M input tokens — 72% less than the flagship Muse Spark. It is also the best capability-per-dollar pick in the lineup.
The safest meta model worth using default, the cheaper option worth trying first, and the specialist pick — before you read the detail below.
Muse Glimmer 30B is the strongest answer here for meta model worth using — pick it when quality of output matters more than the $0.35/1M/1M input you pay for it.
Llama 4 Scout handles the same job for about 8% less per token. Start here and only move up if the output is not good enough.
Llama 4 Maverick is the fastest of these for meta model worth using — worth it when latency is what the reader notices, not the last few points of reasoning depth.
Muse Glimmer 30B is the lowest-cost Meta model: $0.35/1M input, $1.5/1M output.
Muse Glimmer 30B is the best capability-per-dollar pick (budget score 90/100).
Muse Spark costs 4x more on input — reserve it for work where quality is the bottleneck.
Choose Muse Glimmer 30B for high-volume, low-stakes tasks like classification, extraction, and drafts.
Choose Muse Glimmer 30B as the everyday default if you want one budget model.
Route only the hardest tasks to Muse Spark — a two-tier setup usually cuts spend 60–80%.
Switch the scoring lens to see whether the meta model worth using answer changes when cost, speed, or long-document depth leads the decision.
Meta / Balanced / Aug 6, 2026
Best-value multimodal agentic model — GPT-5.5-tier smarts, video/audio/PDF in.
Ranks models by the broadest mix of coding, writing, research, and long-context usefulness.
You need frontier-ceiling reasoning or a mature developer ecosystem — Opus 5 and GPT-5.6 lead both.
Where the meta model worth using recommendation shifts once you weigh price or latency differently.
Apache 2.0 agent model that runs on a 24GB GPU.
Best open-weight long-context option for self-hosted pipelines.
Best flexible option for teams that need open-weight portability.
Best-value multimodal agentic model — GPT-5.5-tier smarts, video/audio/PDF in.
List prices and published scores — the numbers this page's pick is built from.
| Model | Input | Output | Est. month | Context | Speed | Coding | Writing | Research |
|---|---|---|---|---|---|---|---|---|
| Muse Glimmer 30BMeta | $0.35/1M | $1.50/1M | $6.50 | 131k tokens | Fast | 80 | 74 | 76 |
| Llama 4 ScoutMeta | $0.50/1M | $1.20/1M | $7.40 | 512k tokens | Fast | 54 | 60 | 78 |
| Llama 4 MaverickMeta | $0.60/1M | $1.60/1M | $9.20 | 256k tokens | Fast | 58 | 66 | 64 |
| Muse SparkMeta | $1.25/1M | $4.25/1M | $21 | 1.0M tokens | Balanced | 89 | 88 | 90 |
Scores out of 100 — how we evaluate models. “Est. month” is 10M in / 2M out at list price: a ceiling, no discounts.
Why each one is on the shortlist for meta model worth using, what it is genuinely good at, and where we would steer you away from it.
Ranked first here for meta model worth using: 80/100 on coding, with the widest margin of anything in this line-up.
Meta's return to genuine open source — a 30B dense model under Apache 2.0, built for always-on agents rather than chat, and the first release from Meta Superintelligence Labs.
You need a large context window or the strongest agent scores at this size — check Qwen's 27B first.
The best Apache 2.0 agent model you can run on consumer hardware right now. Pick it when licence freedom and local execution matter more than the last few benchmark points — otherwise Qwen3.6-27B edges it on agent tasks.
Full pricing, benchmark table and release notes on the Muse Glimmer 30B page.
The cost-conscious pick for meta model worth using, about 8% less per token than Muse Glimmer 30B than the top choice while holding 54/100 on coding.
Long-window open-weight model that handles large document sets at a low price point.
You want a hosted solution — Gemini 3.1 Flash gives more context for roughly the same cost.
A compelling pick for self-hosted long-context pipelines — but Gemini 3.1 Flash now offers 1M context hosted at a similar price.
Full pricing, benchmark table and release notes on the Llama 4 Scout page.
Here for latency: it answers fastest of anything listed for meta model worth using, at 58/100 on coding.
Best flexible option for teams that need open-weight portability. Full Llama 4 Maverick review →
Also worth a look for meta model worth using, at 89/100 on the coding axis.
Best-value multimodal agentic model — GPT-5.5-tier smarts, video/audio/PDF in. Full Muse Spark review →
UseRightAI recommendations are based on practical decision factors people actually feel in day-to-day use.
Newsletter
We email when the meta model worth using pick changes, when one of these models moves on price, or when something new displaces the current leader.
No spam. Useful updates only. Affiliate disclosures always clearly labeled.
Muse Glimmer 30B at $0.35/1M input and $1.5/1M output tokens. Apache 2.0 agent model that runs on a 24GB GPU.
Muse Glimmer 30B is the best capability-per-dollar pick in Meta's lineup (budget score 90/100). It handles local and self-hosted agents that run continuously well — step up to Muse Spark only where quality visibly falls short.
Muse Glimmer 30B costs $0.35/1M input vs $1.25/1M for Muse Spark — a 72% saving on input tokens.
Llama 4 Scout — 512K tokens at $0.5/1M input. Context is where budget models are least compromised: you usually lose reasoning depth before you lose window size, so a cheap model is often a perfectly good choice for summarising or extracting from long documents.
Qwen3.6-27B beats it on several practical agent and multimodal tests in Meta's own comparison table. Meta publishes no first-party API price — you pay a third-party host or run it yourself. Avoid it if you need a large context window or the strongest agent scores at this size — check Qwen's 27B first.
On a moderate workload of 10M input and 2M output tokens, Muse Glimmer 30B runs about $6.50 against $21.00 for Muse Spark — a difference of $14.50 a month at the same volume. Output tokens dominate the bill on both, so the length of the responses you generate matters far more than the length of your prompts.
Mixing is almost always cheaper for the same quality. Route high-volume, low-stakes work — classification, extraction, first drafts, routine agent steps — to Muse Glimmer 30B, and reserve Muse Spark for the calls where a wrong answer costs real time. Teams that split this way typically cut spend substantially without a quality drop anyone notices, because most tokens in a real workload are not hard problems.