Claude Fable 5 vs Claude Haiku 4.5
Select two models to compare their coding performance, pricing, and features.
Key Takeaways
| Metric | Claude Fable 5 | Claude Haiku 4.5 |
|---|---|---|
| Coding Index | 114.18 | — Score awaiting verification |
| Coding tier | S | — |
| Input Price | $10/1M | $1/1M |
| Output Price | $50/1M | $5/1M |
| Context Window | 1M | 200K |
DeepSeek V4 Flash 0731 = 100. The index reflects two shared tests, not capability multiples or statistically established gaps.
Verdict
Models awaiting tier review remain in the catalog. An unassigned tier does not mean weaker capability.
Capabilities & Use Cases Comparison
Claude Fable 5 Tier S
- •Complex architecture
- •Multi-file refactoring
- •Claude Code
- •Code review
- •Highest coding benchmark score
- •Adaptive reasoning with Opus fallback
- •Top-tier architecture design
- •Best for complex coding tasks
Claude Haiku 4.5 Tier —
- •Quick completions
- •Simple functions
- •Autocomplete
- •High-volume tasks
- •Fastest Claude model for coding
- •Great value for money
- •Good for quick code completions
- •Low latency responses
Claude Fable 5
Benchmark evidence and ranking method · Claude Fable 5
Cross-checked
Terminal-Bench 4.0 · Claude Code · max: 44.55% · 95% CI ± 3.85%
Artificial Analysis · Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)
Test configuration: max
Terminal-Bench 2.1: 84.64% · SciCode: 61.00%
LiveBench · 2026-06-25 · claude-fable-5-max-effort
Coding: 85.99% · Agentic Coding: 62.17%
ModelRank assigns editorial tiers using public coding evidence, not automatic index thresholds or statistically proven classes. Price, speed, context size and sponsorship do not affect tiers.
DeepSeek V4 Flash 0731 = 100. The index reflects two shared tests, not capability multiples or statistically established gaps.
Results reflect particular model configurations. Agent harnesses, reasoning effort, tools, budgets and tasks may differ. Close indices do not establish superiority. B describes a relative position in this selected group, not an inability to code.
Data checked: 2026-09-10. Snapshot date is the retrieval date, not necessarily the evaluation date.
Claude Haiku 4.5
Results reflect particular model configurations. Agent harnesses, reasoning effort, tools, budgets and tasks may differ. Close indices do not establish superiority. B describes a relative position in this selected group, not an inability to code.
API cost estimate
1M input + 200K output (USD):
Claude Fable 5: $20.0000
Claude Haiku 4.5: $2.0000
Estimates use the displayed channel and time band; DeepSeek uses peak rates. Cache discounts, tool fees, long-context price tiers and uncounted additional tokens are excluded.
Frequently Asked Questions
Which model is better for coding: Claude Fable 5 or Claude Haiku 4.5?
A comparable reference index is unavailable for one or both models; no index-based winner is assigned. Editorial tiers: Claude Fable 5 = S; Claude Haiku 4.5 = No tier assessment yet. Tiers are separate editorial assessments and do not establish a reliable order within a tier.
How do Claude Fable 5 and Claude Haiku 4.5 compare in API pricing and costs?
Claude Fable 5: $10/1M input, $50/1M output (OpenRouter quote); Claude Haiku 4.5: $1/1M input, $5/1M output (OpenRouter quote). 1M input + 200K output (USD): $20.0000 vs $2.0000.
What are the context window sizes of Claude Fable 5 and Claude Haiku 4.5?
Claude Fable 5 provides a 1,000,000 token context window, while Claude Haiku 4.5 provides 200,000 tokens.