Models
July 5, 2026 · View on GitHub
The current Claude lineup at a glance, plus deeper specs for the headline models. For Anthropic's authoritative list (always more current), see the official model overview.
The current lineup (July 2026)
| Model | Model ID | Context | Max output | Best for |
|---|---|---|---|---|
| Sonnet 5 (default) | claude-sonnet-5 | 1M | 128K | Everyday coding — most tasks live here |
| Opus 4.8 (Opus flagship) | claude-opus-4-8 | 1M | 128K | Complex reasoning, large refactors, agent orchestration |
| Fable 5 (Mythos class) | claude-fable-5 | 1M | 128K | The genuinely hard problems — a tier above Opus |
| Haiku 4.5 (fast/cheap) | claude-haiku-4-5 | 200K | 64K | Quick edits, doc updates, light reasoning |
⚠️ Anthropic ships new models often. Always check the official model overview before pinning a model ID.
1M context is standard on all current Opus/Sonnet/Fable models — no beta flag, and the full window bills at standard per-token rates (no long-context surcharge).
Legacy models still available: Opus 4.7, Opus 4.6, and Sonnet 4.6 remain accessible via API and the /model switcher — useful for pinning a build to a specific behavior. Opus 4.1 is deprecated and retires August 5, 2026.
Mythos 5 (claude-mythos-5) is the same underlying model as Fable 5 with safeguards lifted in some areas — invitation-only for approved organizations via Project Glasswing, no self-serve access. See the announcement.
The headliners
Claude Sonnet 5 — the new default
Announced June 30, 2026; Claude Code's default model since v2.1.197. The default for Free and Pro on claude.ai too.
- 1M-token context, 128K max output — whole-codebase work without splitting.
- Intro pricing: $2/$10 per MTok through August 31, 2026, then $3/$15.
- Defaults to
higheffort on API and Claude Code.
Claude Opus 4.8 — the Opus flagship
Announced May 28, 2026, replacing Opus 4.7 at unchanged pricing ($5/$25).
- Launched alongside Dynamic Workflows (research preview) — orchestrating hundreds of parallel subagents.
- Anthropic reports it is ~4× less likely than Opus 4.7 to let flaws in its own code pass unflagged.
- Defaults to
higheffort on all surfaces; supportsxhigh. - The default model for Fast Mode (see below).
Claude Fable 5 — the Mythos class
Announced June 9, 2026 with Mythos 5 — the first Mythos-class models, a capability tier above Opus.
- $10/$50 per MTok (batch $5/$25).
- Adaptive thinking is always on —
thinking: {type: "disabled"}is rejected. - Dual-use safety classifiers (cyber, bio/chem, distillation) fall back to Opus 4.8 instead of refusing.
- Uses the Opus 4.7 tokenizer — ~30% more tokens for the same text vs pre-4.7 models; budget accordingly.
- Was included free on Pro/Max/Team from June 9–22, 2026; since June 23 it draws usage credits on subscription plans. GA on API, Bedrock, Google Cloud, and Foundry.
Fast Mode
Fast Mode delivers up to 2.5× faster output at 2× the standard price, running on Opus 4.8 (the default fast-mode model since Claude Code v2.1.154). Toggle with /fast (CLI-only, v2.1.36+); the ↯ indicator confirms it's on. On subscription plans it draws from usage credits.
| Standard Opus 4.8 | Fast Mode (Opus 4.8) | |
|---|---|---|
| Input (per MTok) | $5 | $10 (2×) |
| Output (per MTok) | $25 | $50 (2×) |
⚠️ Older Opus fast modes are being retired: Opus 4.7 fast ($30/$150) was deprecated June 25, 2026 and is removed July 24, 2026 (requests will error, no fallback). Opus 4.6 has silently run at standard speed since June 29, 2026.
Pricing
Subscription plans
| Plan | Price | Models | Usage |
|---|---|---|---|
| Pro | $20/mo ($17/mo annual) | All current | Base — five-hour limits doubled May 6, 2026 |
| Max 5x | from $100/mo | All current | 5× Pro |
| Max 20x | $200/mo | All current | 20× Pro |
API (pay-as-you-go)
| Model | Input (per MTok) | Output (per MTok) |
|---|---|---|
| Sonnet 5 | $2 (intro, → $3 after Aug 31, 2026) | $10 (intro, → $15) |
| Opus 4.8 | $5 | $25 |
| Fable 5 | $10 | $50 |
| Haiku 4.5 | $1 | $5 |
- Fable 5 batch: $5/$25. Cache pricing: $12.50 (5-min write) / $20 (1-hr write) / $1 (read) per MTok.
- The Batch API supports 300K output tokens on Opus 4.8/4.7/4.6 and Sonnet 5/4.6 via the
output-300k-2026-03-24beta header (model overview).
Pricing changes occasionally. Check claude.com/pricing and the API pricing docs before committing.