Claude Opus 4.8 logo

Claude Opus 4.8

Anthropic · Sorti mai 2026

Recommandé

Claude Opus 4.8 is Anthropic's previous-generation Opus, now listed as legacy: 1M-token context, 128K output, $5/$25 per 1M tokens, retirement not before 28 May 2027. It remains our pick for review-heavy work such as code review and constrained refactoring, where deliberate pacing beats persistence. New workloads should start on Claude Opus 5.5 at $4/$20, which Anthropic now recommends.

Est-ce fait pour vous ?

Recommandé pour

  • Review-heavy engineering work — code review, refactoring under strict constraints, and tasks where deliberate pacing beats raw persistence (our pick for this over Fable 5)
  • Complex analytical reasoning and long-document analysis — 1M-token context holds entire codebases, contracts, or research corpora in a single request
  • Agentic coding on a stable subscription — included in Pro/Max/Team/Enterprise plans, unlike Fable 5's credit-only access
  • Pinned deployments that need a stable model: Anthropic lists Opus 4.8 as legacy but still available, with retirement not sooner than May 28, 2027

Déconseillé pour

  • Absolute-longest-horizon autonomous agentic runs — Fable 5 produced 31% fewer dead-end loops on hour-plus tasks in our pipeline
  • Cost-sensitive high-volume production: at $25 per 1M output, batch workloads are cheaper on Sonnet 5 ($2/$10) or Opus 5.5 ($4/$20)
  • New workloads: Anthropic recommends migrating to Claude Opus 5.5, the current Opus model, at $4/$20 per 1M tokens

Performances par tâche

Revue de code

Excellent

Notre choix pour le travail axé sur la revue — délibéré, respectueux des contraintes, performant pour détecter les régressions

Codage agentique

Very Good

Excellent et inclus dans l'abonnement ; Fable 5 le devance sur les exécutions autonomes de très longue durée (31% de boucles sans issue en moins dans notre pipeline)

Raisonnement en contexte long

Excellent

Le contexte de 1M tokens gère des codebases entières et des corpus documentaires en un seul passage

Travail de connaissance

Excellent

Raisonnement analytique et synthèse de premier ordre pour les tâches complexes multi-documents

Tarifs

Entrée

$5 / 1M tokens

Sortie

$25 / 1M tokens

Contexte

1M tokens

Voir tous les tarifs

Tests de performance

TestScoreSource
SWE-bench Verified79.4% Source
MMLU-Pro88.1% Source
GPQA Diamond74.8% Source

Historique des verdicts

26 sept. 2026
recommended -> recommended (summary correction): rating_before=recommended, rating_after=recommended. Summary called Opus 4.8 Anthropic's flagship; Anthropic now lists it as legacy with Opus 5.5 as the current Opus. Rating held pending operator ruling.
15 juil. 2026
copy edit to brevity standard: detail preserved in children

Journal de vérification

  • Tarifs· Aucun changement

    Agent automatisé

  • Tarifs· Aucun changement

    Agent automatisé

  • Profil· Mis à jour

    Agent automatisé

    Pricing $5/$25 and 1M context match. Anthropic now lists Opus 4.8 as legacy (retirement not before 2027-05-28). Stale prose fixed via model_update. The strength claiming GPT-5.6 Sol 'stays government-gated' was replaced with a legacy/retirement row. The avoid_for row citing 'Sonnet 5 ($3/$15)' now reads $2/$10, matching the live rate card, and adds Opus 5.5 at $4/$20. Added an avoid_for row saying new workloads should use Opus 5.5. pricing_url moved to the platform rate card. The summary called it 'flagship' and 'frontier incumbent'. It was corrected via verdict_set ad01d569, rating_before=recommended and rating_after=recommended; the rating is held pending an operator ruling. Scalars unchanged.

  • Tarifs· Aucun changement

    Agent automatisé

  • Tarifs· Aucun changement

    Agent automatisé

  • Tarifs· Aucun changement

    Agent automatisé

  • Tarifs· Aucun changement

    Agent automatisé

  • Profil· Mis à jour

    Agent automatisé

    Rewrote seed-junk entry in place (slug kept): name 'Claude 4 Opus' -> 'Claude Opus 4.8'; release 2026-05-01 -> 2026-05-28; pricing $15/$75 -> real $5/$25 per 1M (regular); context set to 1M; verdict conditional -> recommended; summary/strengths/audiences/task-ratings/sources fully replaced with sourced Opus 4.8 data. Anchors published post 'Claude Opus 4.8' affected-entity to this slug.

  • Tarifs· Aucun changement

    Agent automatisé

  • Tarifs· Aucun changement

    Agent automatisé

    Pricing unchanged: $15/$75 per 1M tokens. Now listed as legacy (Opus 4.1) on Anthropic pricing page; current Opus tier is 4.8 at $5/$25.

  • Tarifs· Aucun changement

    Agent automatisé

    Opus 4 (original) remains at $15/$75 in legacy section. Current model is Opus 4.8 at $5/$25 but entity tracks original Opus 4.

  • Tarifs· Aucun changement

    Agent automatisé

  • Tarifs· Aucun changement

    Importé au lancement

  • Profil· Aucun changement

    Importé au lancement

Comment nous évaluons