TL;DR: Claude has four tiers in mid-2026. Haiku 4.5 is fast and cheap ($1/$5 per million tokens). Sonnet 4.6 is the balance pick ($3/$15). Opus 4.8 is the most capable ($5/$25). Fable 5 sits on top ($10/$50).
Anthropic ships Claude in a small, clear lineup. There are three named families plus one new top tier. Haiku is the fast, low-cost model. Sonnet is the everyday balance of speed and smarts. Opus is the most capable model for hard, long tasks. And Fable 5, released above Opus, is the most powerful model Anthropic offers. The current versions are Claude Haiku 4.5, Claude Sonnet 4.6, Claude Opus 4.8, and Claude Fable 5.
Key takeaways
- Four tiers in mid-2026: Haiku (fast), Sonnet (balanced), Opus (most capable), Fable 5 (most powerful).
- Output is the price driver. Fable 5 output costs $50 per million, ten times Haiku at $5.
- Sonnet, Opus, and Fable 5 all have a 1M token context window. Haiku has 200K.
- Adaptive thinking replaced fixed thinking budgets. You set effort from low to max instead.
- Most teams should default to Sonnet 4.6 and only reach for Opus or Fable 5 when the task is genuinely hard.
The current Claude models at a glance
Here is the full lineup of current models, with the exact model ID your engineers will type into the API. Use this table as your reference. Everything below explains the details.
| Model | Model ID | Context | Max output | Price in / out (per 1M) | Best for |
|---|---|---|---|---|---|
| Fable 5 | claude-fable-5 | 1M | 128K | $10 / $50 | The hardest reasoning and the highest stakes work |
| Opus 4.8 | claude-opus-4-8 | 1M | 128K | $5 / $25 | Long agentic work, big refactors, deep research |
| Opus 4.7 | claude-opus-4-7 | 1M | 128K | $5 / $25 | Previous Opus, still strong and still active |
| Sonnet 4.6 | claude-sonnet-4-6 | 1M | 64K | $3 / $15 | The default for most production work |
| Haiku 4.5 | claude-haiku-4-5 | 200K | 64K | $1 / $5 | Fast, cheap, high-volume simple tasks |
Which Claude model fits your team?
Pick your main use case to see the right tier.
For chatbots, tagging, and routing at scale, Haiku is fast and costs $1/$5 per million tokens. You only pay for Opus when a task truly needs it. See the cost of building this in-house on our developer rate card.
For everyday coding, reviews, and tool use, Sonnet is the balance pick at $3/$15. Pair it with engineers who know Claude Code. Hire AI agent developers who build with it daily.
For week-long refactors, deep research, and long autonomous runs, the top tiers earn their price. Strong AI and machine learning engineers get the most from them.
The model is the easy part. The hard part is the team that builds and ships with it. We match you with pre-vetted engineers in about 24 hours. Tell us what you need.
The Claude family
Anthropic keeps the naming simple. Each family is named after a form of writing. The bigger the form, the bigger the model. Haiku is short. Sonnet is mid-length. Opus is a large work. Fable 5 broke the pattern as a new top tier above Opus.
Claude Haiku 4.5
Haiku is the fastest and cheapest model. It has a 200K token context window and 64K max output. Use it for simple, high-volume tasks. Think classification, tagging, routing, short replies, and first-pass drafts. At $1 input and $5 output per million tokens, you can run a lot of traffic through it without a large bill.
Claude Sonnet 4.6
Sonnet is the best mix of speed and intelligence. It has a 1M token context window and 64K max output. It costs $3 input and $15 output per million tokens. For most production work, Sonnet is the right default. It handles coding, reviews, summaries, extraction, and tool use well. We tell most clients to start here and move up only when they hit a real limit.
Claude Opus 4.8
Opus is the most capable Opus-tier model. It is highly autonomous and strong on long, multi-step work that runs for hours. Big code migrations, deep research, and agents that keep their own memory are where it shines. It has a 1M context window and 128K max output, at $5 input and $25 output per million tokens. Opus 4.7 is the previous version and is still active if you need to pin it.
Claude Fable 5
Fable 5 is the most powerful and most intelligent model Anthropic offers. It is a new tier that sits above Opus. It shares the same 1M context window and 128K max output as Opus 4.8. The price is higher, at $10 input and $50 output per million tokens. Reach for Fable 5 when the task is the hardest you have and the cost of a wrong answer is high. For most teams, this is a small share of total work.
What every current model can do
The four tiers share the same core skills. The difference is depth and speed, not a missing feature. Every current model can read images, call tools, and return strict JSON when you ask for it. That last one, called structured outputs, forces the reply to match a schema you set, which makes the data safe to parse in code.
Vision is built in. The model reads screenshots, charts, PDFs, and photos. Opus 4.7 and newer added high-resolution image support, which helps with detailed screens and document work. Tool use lets the model call your functions, run code in a sandbox, search the web, and read files. This is what powers agents that do real work, not just chat.
Two features save real money at scale. Prompt caching stores a large fixed prompt once and reuses it cheaply across many calls. Batch processing runs non-urgent jobs at half price. A team that uses both can serve heavy workloads for far less than the sticker rate suggests. Knowing which feature to reach for is part of what we test when we vet engineers.
Pricing compared across the lineup
Output tokens are the real cost driver. Most apps read a prompt and write a longer answer, so output volume adds up fast. The gap between tiers is large. Haiku output is $5 per million. Fable 5 output is $50 per million, which is ten times more. The chart below shows output price for each current model.
One way to cut cost is to route by difficulty. Send easy requests to Haiku and only escalate the hard ones to Opus or Fable 5. Another lever is prompt caching, which charges cached context at about a tenth of the input price. Teams that build this routing layer well often cut their model bill by half or more. This is the kind of work that separates a strong AI engineer from an average one.
Context windows and output limits
The context window is how much text the model can read at once, counting both your prompt and the conversation so far. A 1M token window is roughly 750,000 words. That is enough to drop a whole codebase or a stack of long documents into a single request.
Sonnet 4.6, Opus 4.8, Opus 4.7, and Fable 5 all have the 1M window. Opus and Fable also price that large context at the standard rate, with no long-context premium. Haiku 4.5 has a smaller 200K window, which is still about 150,000 words and fine for most chat and tagging work. The chart compares context size and max output across the lineup.
Max output is a separate limit. It caps how long a single reply can be. Haiku and Sonnet top out at 64K tokens. Opus and Fable 5 reach 128K. For very long outputs you must stream the response, or the request can time out. Your engineers should know this. It is a common cause of broken long-generation jobs.
How thinking and effort work
Older Claude models used a fixed thinking budget. You set a token count and the model thought up to that limit. The current models replaced that with adaptive thinking. The model decides on its own when to think and how deeply, based on the task. You no longer set a token budget for thinking.
You steer cost and depth with an effort setting instead. The levels run from low to medium to high, plus xhigh and max on the top models. Low effort means fewer tool calls and shorter answers, which is good for cheap, fast tasks. High effort means deeper reasoning, which fits hard coding and agent work. The newest models respect effort more strictly, so set it on purpose rather than leaving the default.
This matters for hiring. An engineer who understands adaptive thinking and effort will tune a workflow to use far fewer tokens at the same quality. We see this gap often when we vet candidates. The ones who have shipped real Claude apps know these levers cold. You can read more in our breakdown of Claude versus ChatGPT for coding.
Which model should you use
Start with Sonnet 4.6. It is the safe default for most teams and most tasks. Move down to Haiku when the task is simple and you run high volume. Move up to Opus or Fable 5 when the task is genuinely hard or the cost of a wrong answer is high. The table below maps common jobs to a tier.
| Job | Recommended tier | Why |
|---|---|---|
| Support chatbot, FAQ | Haiku 4.5 | Fast and cheap at high volume |
| Data tagging, routing | Haiku 4.5 | Simple, repeated, low risk |
| Daily coding and code review | Sonnet 4.6 | Strong balance of speed and skill |
| Document summaries, extraction | Sonnet 4.6 | 1M context, good accuracy, fair price |
| Large refactor, long agent run | Opus 4.8 | Highly autonomous over long tasks |
| Hardest reasoning, high stakes | Fable 5 | Most powerful, worth the premium |
A simple rule helps. If a cheaper model gives the right answer most of the time, use it and add a check. Escalate only the cases that fail. This keeps quality high and cost low. We worked with a fintech team that ran every request on Opus. Moving routine checks to Sonnet and tagging to Haiku cut their bill by more than half with no drop in quality.
Legacy and retired models
Anthropic keeps older models running for a while, then retires them. If your code still calls an old model ID, plan to move it. The table shows the main older models and what to use instead. Retired models return an error, so those need a fast fix.
| Older model | Status | Move to |
|---|---|---|
| Opus 4.6, Opus 4.5, Opus 4.1 | Still active | Opus 4.8 |
| Sonnet 4.5, Sonnet 4 | Active or deprecated | Sonnet 4.6 |
| Haiku 3.5 | Retired | Haiku 4.5 |
| Sonnet 3.7, Sonnet 3.5 | Retired | Sonnet 4.6 |
| Opus 3 | Retired | Opus 4.8 |
A move is usually small. Change the model ID, switch any fixed thinking budget to adaptive thinking, and remove old sampling settings that the new models no longer accept. The current Anthropic models overview lists every active and retired ID in one place, so check it before you migrate.
What this means for hiring
Knowing the model lineup is step one. Getting value from it is the harder step. The teams that win are not the ones with the biggest model. They are the ones who route work to the right tier, cache context, set effort well, and ship a clean product around it. That is engineering, not just a model choice.
Strong AI engineers are scarce and costly in the United States. A senior AI engineer there runs $12,000 to $18,000 per month all in. The same skill from a pre-vetted engineer in Vietnam or the Philippines often costs $3,000 to $6,000 per month. The work is the same. The bill is not. Many of our clients build their whole Claude stack with engineers from our network.
If you want to go deeper on the wider picture, our Claude AI statistics for 2026 covers adoption, revenue, and market share. Read that next to see how fast this market is moving and why the right team matters more each quarter.
Compare other model families
Build your Claude stack with the right team
Claude gives you four tiers and clear prices. The model is the easy part. The team that builds with it is what ships value. We match you with pre-vetted senior engineers across Asia in about 24 hours, with no upfront cost and payroll handled.
