A New Tier Above Opus
For three years, "Opus" meant the top of Anthropic's lineup. That changed with Claude Fable 5 (API model ID claude-fable-5), which launches at $10 per million input tokens and $50 per million output tokens — exactly double Claude Opus 4.8's $5/$25. It ships with a 1M-token context window and Anthropic's adaptive thinking as the only reasoning mode.
The Full Claude Lineup, June 2026
| Model | Input $/1M | Output $/1M | Context |
|---|---|---|---|
| Claude Fable 5 | $10.00 | $50.00 | 1M |
| Claude Opus 4.8 | $5.00 | $25.00 | 1M |
| Claude Opus 4.7 | $5.00 | $25.00 | 1M |
| Claude Opus 4.6 | $5.00 | $25.00 | 1M |
| Claude Sonnet 4.6 | $3.00 | $15.00 | 1M |
| Claude Haiku 4.5 | $1.00 | $5.00 | 200K |
Note that three generations of Opus remain available at identical pricing — useful if your prompts are tuned to a specific version and you want to migrate on your own schedule.
What Does a Real Workload Cost?
A coding agent that processes 5M input and 1M output tokens per day costs about $100/day on Fable 5 versus $50/day on Opus 4.8. A customer-support assistant handling 50K conversations a month (roughly 100M input / 25M output tokens) would run about $2,250/month on Fable 5 against $1,125 on Opus 4.8.
When Is 2x the Price Worth It?
Per-token price is the wrong lens for agentic work. If Fable 5 completes a long-horizon task in one pass that Opus needs two attempts at — or that needs an engineer to intervene — the premium pays for itself immediately. Early reports point to its biggest gains in exactly those areas: multi-hour autonomous coding runs, complex refactors, and deep research tasks.
For high-volume, well-bounded tasks — classification, extraction, summarization, routine chat — stay on Sonnet 4.6 or Haiku 4.5. Paying flagship rates for work a $1/$5 model handles correctly is the most common AI budget mistake we see.
Compare It Yourself
See how Fable 5 stacks up against GPT-5.5, Gemini 3.1 Pro, and Grok 4.3 in our model comparison tool, or estimate your exact monthly spend with the cost calculator. For the agentic-workload view, our Fable 5 vs GPT-5.5 guide runs the head-to-head numbers.