Claude Fable 5.2 vs Claude Haiku 5.5: Specs & Benchmark Comparison

Claude Fable 5.2 and Claude Haiku 5.5 both come from Anthropic, Claude Haiku 5.5 was released in October 2026, and Claude Fable 5.2 followed a month later in November 2026.

The two models share 22 published benchmarks. Claude Fable 5.2 leads on 21 of them, Claude Haiku 5.5 on 1. The widest gaps are on Terminal-Bench-Science 0.1, where Claude Fable 5.2 scores 74.6% against 20.6%; SWE-Bench Multimodal, where Claude Fable 5.2 scores 74.1% against 30.7%. Averaged across everything we track, Claude Fable 5.2 sits at 73.1% and Claude Haiku 5.5 at 57.7%.

Claude Haiku 5.5 is the cheaper API at $0.10 per million input tokens and $0.50 per million output tokens, roughly 80 times cheaper than Claude Fable 5.2 at $8 and $40. Both accept a context window of about 1M tokens. Both accept text and images as input.

CharacteristicClaude Fable 5.2Claude Haiku 5.5
CompanyAnthropicAnthropic
Release DateNovember 9, 2026October 7, 2026
Parameters——
MultimodalYesYes
Context (input)1.0M1.0M
Context (output)128K128K
Input Price / 1M$8.00$0.10
Output Price / 1M$40.00$0.50
Average Score73.1%57.7%
Benchmarks
Terminal-Bench-Science 0.174.6%20.6%
SWE-Bench Multimodal74.1%30.7%
Terminal-Bench 4.067.6%39.2%
SWE-Bench Pro93.0%64.8%
Chartography (no tools)73.4%46.4%
FrontierSWE V269.9%43.8%
PhysicianBench64.4%43.0%
OSWorld 2.1 (strict)54.6%37.1%
OSWorld 2.1 (partial)89.3%72.4%
Humanity's Last Exam (with tools)74.2%57.4%
SWE-bench Multilingual96.6%83.7%
OfficeQA Pro72.9%60.3%
OfficeQA83.4%73.5%
Global-MMLU96.0%87.8%
GDPval-AA 2.161.0%54.0%
AA-Briefcase v1.159.5%52.6%
MILU94.1%87.6%
Chartography91.6%86.2%
Program Bench85.6%82.0%
HealthBench Professional67.7%64.8%
HealthBench59.6%61.6%
HealthBench Professional (raw)71.1%71.0%

Visual Benchmark Comparison

Claude Fable 5.2
Claude Haiku 5.5
Terminal-Bench-Science 0.10.7 vs 0.2
0.7
0.2
SWE-Bench Multimodal0.7 vs 0.3
0.7
0.3
Terminal-Bench 4.00.7 vs 0.4
0.7
0.4
SWE-Bench Pro0.9 vs 0.6
0.9
0.6
Chartography (no tools)0.7 vs 0.5
0.7
0.5
FrontierSWE V20.7 vs 0.4
0.7
0.4
PhysicianBench0.6 vs 0.4
0.6
0.4
OSWorld 2.1 (strict)0.5 vs 0.4
0.5
0.4
OSWorld 2.1 (partial)0.9 vs 0.7
0.9
0.7
Humanity's Last Exam (with tools)0.7 vs 0.6
0.7
0.6
SWE-bench Multilingual1.0 vs 0.8
1.0
0.8
OfficeQA Pro0.7 vs 0.6
0.7
0.6
OfficeQA0.8 vs 0.7
0.8
0.7
Global-MMLU1.0 vs 0.9
1.0
0.9
GDPval-AA 2.10.6 vs 0.5
0.6
0.5

Verdict

Both models show equal results — the choice depends on your specific use case.

Overall Performance

Both models show comparable average scores: Claude Fable 5.2 — 0.7, Claude Haiku 5.5 — 0.6.

API Cost

Claude Haiku 5.5 is 80.0x cheaper: input $0.10/1M vs $8.00/1M tokens.

Context Window

Same context size: 1M tokens.

Recency

Claude Fable 5.2 is newer: released 11/9/2026 vs 10/7/2026.

More About These Models

Related Comparisons

Frequently Asked Questions

Which is better for coding — Claude Fable 5.2 or Claude Haiku 5.5?
Direct comparison on the SWE-Bench benchmark is not available. We recommend reviewing other metrics on the comparison page.
Which model is cheaper — Claude Fable 5.2 or Claude Haiku 5.5?
Claude Haiku 5.5 is cheaper for input: $0.10 per 1M tokens vs $8.00.
Which has a larger context window — Claude Fable 5.2 or Claude Haiku 5.5?
Claude Fable 5.2 supports a larger context: 1,048,576 tokens vs 1,048,576.

The Claude Fable 5.2 and Claude Haiku 5.5 comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the Claude Fable 5.2 or Claude Haiku 5.5 page. See also the complete list of AI model comparisons.