Claude Models Explained
Claude model choice is a capability, cost, and latency decision. As of July 6, 2026, the official docs name Fable 5, Opus 4.8, Sonnet 5, and Haiku 4.5 as the current comparison set, with Mythos 5 in limited availability.
Anthropic’s current model comparison centers on Fable, Opus, Sonnet, and Haiku, with Mythos treated separately for limited trusted access.
Source: Models overview Citable fact $2 / $10 MTokSonnet 5 launched with introductory API pricing of $2 per million input tokens and $10 per million output tokens through August 31, 2026.
Source: Introducing Claude Sonnet 5The Current Lineup
The official Models overview describes Claude as a family of state-of-the-art large language models and lists the latest comparison across Claude Fable 5, Claude Opus 4.8, Claude Sonnet 5, and Claude Haiku 4.5. It also notes that current Claude models support text and image input, text output, multilingual capabilities, and vision.
The practical difference is not just benchmark rank. Fable is the highest-capability general release, Opus is the premium agentic and enterprise workhorse, Sonnet is the broad default for cost-performance, and Haiku is tuned for speed and high-volume economics.
Claude Fable 5
Highest broadly released capability, especially for long-running and complex work; use where cost and safeguards are acceptable.
Claude Opus 4.8
Premium reasoning and agentic coding model; good default for complex enterprise or ambiguous tool-heavy work.
Claude Sonnet 5
High-performance model for coding, agents, and professional work at lower cost than Opus.
Claude Haiku 4.5
Fastest and most economical current tier for classification, extraction, routing, and high-volume simple tasks.
How To Choose
Choose by failure cost first. If a wrong or incomplete answer is expensive, start high and measure down. If latency or unit economics dominate, start with Haiku or Sonnet and only escalate difficult turns. If the task spans tools, code, browsing, or uncertain state, Opus and Fable are better candidates.
For developers, model choice should be tested against real evals rather than vibes. Use the same prompt, tools, context size, retry rules, and stop handling that production will use. A model that is cheaper per token can still be more expensive if it retries, over-searches, or needs human repair.
Availability And Naming
The model page lists API IDs such as `claude-fable-5`, `claude-opus-4-8`, `claude-sonnet-5`, and `claude-haiku-4-5`. It also says models are available through the Claude API and major cloud platforms, though exact feature support can vary by platform.
Mythos 5 is not a normal public tier. Anthropic’s Fable/Mythos announcement describes Mythos 5 as the same underlying model as Fable 5 with different safeguards for a small trusted group, initially through Project Glasswing. Treat it as a limited-access security capability, not a general model choice.
Pricing Implications
The official pricing page lists Fable 5 at $10 input and $50 output per million tokens, Opus 4.8 at $5 and $25, Sonnet 5 at introductory $2 and $10 through August 31, 2026, then $3 and $15, and Haiku 4.5 at $1 and $5. Prompt caching changes the economics for repeated context, and batch processing can reduce costs for asynchronous workloads.
For consumer plans, model access is constrained by plan, capacity, and product policy rather than raw API pricing. For API work, output tokens usually dominate cost, so concise output schemas and structured tool results matter.
If you are unsure which model to use, start with Claude Opus 4.8.
Cite This Page
Claude Central. "Claude Models Explained." Last checked July 6, 2026. https://claudecentral.com/models
FAQ
What is the best Claude model?
There is no universal best. Fable 5 is the highest-capability broadly released tier, Opus 4.8 is the strong premium default for complex agentic work, Sonnet 5 is the broad cost-performance choice, and Haiku 4.5 is the speed and cost choice.
Is Claude Mythos available to everyone?
No. Anthropic describes Mythos 5 as limited availability for approved customers and trusted access paths, not a general public model.
Do all models support images?
The official Models overview states that current Claude models support text and image input, text output, multilingual capabilities, and vision.