Claude Models Explained (2026): Opus vs Sonnet vs Haiku vs Fable
Claude comes in a small family of models that trade off intelligence, speed, and cost: Fable 5 is the most capable, Opus 4.8 is the flagship for hard reasoning, Sonnet 4.6 is the everyday balance of smart-and-fast, and Haiku 4.5 is the cheapest and quickest. The naming sounds confusing, but the logic is simple — bigger names cost more and think harder; smaller names are faster and cheaper.
This guide lays out the full 2026 lineup, what each model is actually good for, how the consumer plans map to them, and how to pick without overthinking it.
The lineup at a glance
| Model | Model ID | Context | Input $/1M | Output $/1M | Notes |
|---|---|---|---|---|---|
| Claude Fable 5 | claude-fable-5 | 1M | $10 | $50 | Most capable widely-released model; GA June 9, 2026 |
| Claude Opus 4.8 | claude-opus-4-8 | 1M | $5 | $25 | Flagship Opus (~May 28, 2026) |
| Claude Opus 4.7 | claude-opus-4-7 | 1M | $5 | $25 | Previous Opus |
| Claude Sonnet 4.6 | claude-sonnet-4-6 | 1M | $3 | $15 | Best speed/intelligence balance |
| Claude Haiku 4.5 | claude-haiku-4-5 | 200K | $1 | $5 | Fastest and cheapest |
The pattern: as you move down the table, models get cheaper and faster but a bit less capable on the hardest tasks. Fable and Opus carry a 1M-token context window; Haiku trades context (200K) for speed and price.
Opus: the flagship for hard problems
Opus 4.8 is the model you reach for when the task is genuinely hard — multi-step reasoning, complex code, dense analysis, or anything where a wrong answer is expensive. It’s the top of the “everyday” tiers (below the newer Fable 5) and carries the full 1M-token context.
Use Opus when: you’re debugging a thorny codebase, reasoning over a long legal or financial document, doing serious research synthesis, or you simply want the strongest answer and don’t mind paying more per token.
Skip Opus when: the task is routine. Paying Opus rates to summarize a short email is overkill — Sonnet or Haiku will do it just as well for a fraction of the cost.
Sonnet: the everyday workhorse
Sonnet 4.6 is the model most people should default to. It hits the best balance of speed and intelligence — fast enough for interactive work, smart enough for the vast majority of writing, coding, and analysis tasks, and priced in the middle at $3/$15 per million tokens. It also carries the 1M-token context.
Use Sonnet when: drafting and editing, everyday coding, summarizing, most chat work, and anything where you want a great answer quickly without flagship pricing. For a lot of users, Sonnet is the only model they ever need.
Haiku: fast and cheap
Haiku 4.5 is the fastest and cheapest model at $1/$5 per million tokens, with a 200K context window. It’s built for volume and latency-sensitive work where you don’t need flagship reasoning.
Use Haiku when: classifying or tagging at scale, simple extraction, quick lookups, high-throughput pipelines, or anything where speed and cost matter more than squeezing out the last few points of quality.
Fable 5: the most capable
Fable 5 (GA June 9, 2026) is the most capable widely-released Claude model — the one to use when you want the absolute best output and the task justifies the premium ($10/$50 per million tokens, 1M context). It sits above Opus on raw capability.
Use Fable 5 when: the stakes are high and quality is non-negotiable — your hardest reasoning, your most demanding writing, the work you’d otherwise hand to your sharpest specialist. For routine tasks the marginal gain over Opus or Sonnet rarely justifies the cost.
How the consumer plans map to models
If you use the Claude app rather than the API, you mostly pick a plan, and the plan decides which models you can run:
- Free — runs Sonnet with a daily message limit. Fine for light use.
- Pro ($20/mo) — runs Sonnet 4.6 with much higher limits. The right tier for most individuals.
- Max ($100/mo and $200/mo) — unlocks the Opus tier (and the highest limits). If you want Opus-class reasoning in the app, you need Max.
So the practical rule on the consumer side: Free and Pro give you Sonnet; Opus needs Max. If you keep hitting walls, see Claude usage limits for how the caps work and when to upgrade.
How to pick (the short version)
- Default to Sonnet 4.6. It covers most work and is the best value for everyday tasks.
- Reach for Opus 4.8 when the task is genuinely hard and a better answer is worth the cost.
- Use Haiku 4.5 for high-volume, latency-sensitive, or simple work.
- Use Fable 5 when only the best will do and the stakes justify the premium.
On the API you choose per call. In the app, you choose a plan — and for serious reasoning, that means Max for Opus.
When you’d rather not pick at all
Choosing models is a developer’s problem. If what you actually want is for the work to get done — your inbox triaged, emails sent, meetings booked, your CRM updated — you don’t want to be tuning model selection at all.
That’s the idea behind Carly, an AI executive assistant that runs frontier models under the hood so you never have to think about Opus vs Sonnet. You tell it in plain English what you want (“set up a system to triage my inbox and book sales calls”), it interviews you, and it builds the workflow — then runs it across Gmail and Outlook, on triggers, 24/7 in the cloud, sending and acting on its own. Each agent even gets its own email address. AI agents start at $35/month, with non-AI steps in a workflow running free and unlimited.
Claude is the engine; Carly is the colleague that uses it. For a direct comparison of the two approaches, see Claude vs Carly.
Frequently Asked Questions
What’s the difference between Opus, Sonnet, and Haiku?
They trade intelligence for speed and cost. Opus 4.8 is the flagship for hard reasoning ($5/$25 per million tokens, 1M context). Sonnet 4.6 is the everyday balance ($3/$15, 1M context). Haiku 4.5 is the fastest and cheapest ($1/$5, 200K context). Above all three sits Fable 5, the most capable model ($10/$50, 1M context).
What is Claude Fable 5?
Fable 5 (GA June 9, 2026) is Anthropic’s most capable widely-released model, sitting above Opus on raw capability. It has a 1M-token context window and is priced at $10/$50 per million tokens. Use it when quality is non-negotiable and the task justifies the premium.
Which Claude model should I use?
Default to Sonnet 4.6 — it handles most writing, coding, and analysis well at a reasonable price. Step up to Opus 4.8 for hard problems, down to Haiku 4.5 for high-volume simple tasks, and up to Fable 5 when only the best output will do.
Which Claude model do I get on the free plan?
The Free plan runs Sonnet with a daily message limit. Pro ($20/mo) gives you Sonnet 4.6 with higher limits, and Max ($100 or $200/mo) unlocks the Opus tier. See Claude usage limits for details on the caps.
How big is the context window?
Fable 5, Opus 4.8, and Sonnet 4.6 all have a 1M-token context window. Haiku 4.5 has 200K. The 1M window lets you work over entire codebases or large document sets in one conversation.
Do I have to pick a model to use Claude for real work?
On the API, yes — you choose per call. In the app, you pick a plan. But if you’d rather not think about it at all, an assistant like Carly runs frontier models for you and just does the work — triaging email, sending, scheduling, and updating tools across Gmail and Outlook. AI agents start at $35/month.
More: Claude vs Carly · Claude vs ChatGPT vs Gemini · Claude usage limits · What is Claude Code · Claude alternatives · Best AI agents for productivity
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
