Muse Spark 1.3 — Specifications
Rédigé par Mustafa Ihsan d’après la documentation officielle du fournisseur · Dernière mise à jour
| Développeur | Meta |
|---|---|
| Type | Raisonnement / agentic |
| Modalité | Text, image, video, audio, PDF → Text |
| Paramètres | Non divulgué |
| Fenêtre de contexte | 1 million |
| Licence | Propriétaire |
| Poids ouverts | Non |
| Publié | 2026-09 |
| Prix de l’entrée | $1.25 /1M |
| Prix de la sortie | $4.25 /1M |
| Fournisseurs d'API | Meta Model API (also in Muse Code) |
What is Muse Spark 1.3?
Muse Spark 1.3 is Meta’s model for agentic and coding work at scale, released on 2 September 2026 on Meta Model API and in Muse Code. l'annonce de Meta says it sustains long, multi-step work better than earlier Muse Spark models, and that in comparisons by Meta engineers it used about 20% fewer tool calls and 25% fewer tokens than Muse Spark 1.2. It accepts text, images, video, audio and PDF, returns text, has a 1,048,576-token context window and supports reasoning effort from minimal to max (default high). Meta’s docs recommend it for new work but warn that audio understanding on 1.3 is not fully supported; use Muse Spark 1.2 for audio.
In Standard mode it costs $1.25 per million input tokens, $0.15 for cached input and $4.25 per million output tokens, with no long-context premium. Contributor mode (muse-spark-1.3-contributor) drops that to $0.10, $0.002 and $0.20, but Meta may then use your prompts, code, files and the model’s responses to train future Meta models, and the max reasoning level is not available. Web search grounding adds $2.50 per 1,000 queries (see Meta’s pricing page).
Muse Spark sits in the middle of Meta’s lineup: Muse Wonder ($4 / $20, in preview) is Meta’s most capable model, and Muse Ember ($0.10 / $0.50 for requests up to 200K input tokens) handles high-volume sub-tasks. Muse Spark Open is a separate open-weight release (816B total, 42B active parameters); Muse Spark 1.3 itself is a hosted model without public weights. Meta’s Model API lists no Llama models, so for open Llama weights see Llama 4 Maverick, and compare monthly costs with the Calculateur de coûts pour les API IA.
Muse Spark 1.3 pricing: API cost per 1M tokens
| Entrée (par million de jetons) | $1.25 |
|---|---|
| Sortie (par million de jetons) | $4.25 |
| Ratio sortie/entrée | 3.4× |
| Moyenne pondérée (4:1 entrée:sortie) | $1.85 par 1 million de jetons |
What Muse Spark 1.3 costs per month
Dépense mensuelle réelle pour un ratio entrée-sortie de 4:1 — le rapport effectivement observé dans une charge de travail type (chat ou RAG).
| Charge de travail | Jetons/mois | Coût par mois |
|---|---|---|
| Projet secondaire | 1 million en entrée / 0,25 million en sortie | $2.31 |
| Petite équipe | 20 millions en entrée / 5 millions en sortie | $46 |
| Production | 200 millions en entrée / 50 millions en sortie | $463 |
Calculez vos propres coûts dans le Calculateur de coûts pour les API IA.
Cheaper alternatives to Muse Spark 1.3
| Modèle | Coût combiné par million de dollars | Vous économisez |
|---|---|---|
| Llama 3.1 8B ouverte | $0.0240 | 99 % moins cher |
| Mistral NeMo 12B ouverte | $0.0292 | 98% cheaper |
| Gemma 3 4B ouverte | $0.0600 | 97 % moins cher |
Questions fréquemment posées
How much does Muse Spark 1.3 cost per 1M tokens?
Muse Spark 1.3 costs $1.25 per 1M input tokens and $4.25 per 1M output tokens. At a typical 4:1 input-to-output mix that blends to about $1.85 per 1M tokens.
How much does Muse Spark 1.3 cost per month?
A small-team workload of 20M input and 5M output tokens a month costs about $46 on Muse Spark 1.3. A side project (1M in / 0.25M out) costs roughly $2.31.
What is a cheaper alternative to Muse Spark 1.3?
Llama 3.1 8B is the strongest cheaper option in our database at $0.0240 per 1M blended — about 99% less than Muse Spark 1.3. It is also open-weight, so self-hosting is an option.
Can I run Muse Spark 1.3 locally?
No. Muse Spark 1.3 is a closed, API-only model — the weights are not released, so it cannot be self-hosted.
Why does Muse Spark 1.3 charge more for output than input?
Output tokens are generated one at a time and cannot be batched the way a prompt can, so they cost the provider more to serve. Muse Spark 1.3 charges 3.4× more for output, which is why prompt-heavy workloads are far cheaper to run than generation-heavy ones.
Les prix indiqués correspondent aux tarifs publics officiels pour l’API principale du modèle et sont mis à jour régulièrement à mesure que les fournisseurs les modifient. Les remises liées aux volumes, aux traitements par lots ou aux entrées mises en cache ne sont pas prises en compte. Comparez tous les modèles côte à côte dans le Base de données des modèles IA ou le Classement des grands modèles linguistiques (LLM).