BlogLLM Benchmark Scores as Published, Side by Side

Claude Fable 5.2: Benchmark Scores as Published

Every result the catalog holds for Claude Fable 5.2: 46 tests in 10 areas, from 5 pages. 41 printed by Anthropic itself, the rest by the company named beside each number. Each number is as printed; nothing is ranked or rescored.

Developer AnthropicReleased 2026-11-09Context window 1MMax output 128KInput Text, imageOutput TextWeights ClosedData version 2026111201, as of 2026-11-12

Knowledge Work

TestValuePrinted bySetup as printed
AutomationBench51.4%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-
OfficeQA Pro72.9%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
GDPval-AA v2.11751Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: xhigh
1831Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
AA-Briefcase v1.11785Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-
OfficeQA83.4%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
Harvey LAB-AA v1.1 (hallucination-gated all-pass rate)7.2%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
AA-Briefcase v1.1 (rubric pass rate)66.1%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-
Harvey LAB-AA v1.1 (tasks meeting every criterion)10.8%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
Remote Labor Index (automation rate)41.3%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more pageEffort: high
Real-World Finance v21637Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
AutomationBench (cost per task)lower is better$4.64Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
$1.3Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: low

Agentic Coding

TestValuePrinted bySetup as printed
SWE-Bench Pro93%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
DeepSWE v1.174.3%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
SWE-bench Multilingual96.6%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
Terminal-Bench 4.067.6%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max; Harness: Claude Code --bare; Fallback: safeguards on; the fallback answered 2.5% of requests, affecting 10% of trials
ProgramBench85.6%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Harness: mini-swe-agent; Context: 1000000
91.6%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Harness: mini-swe-agent with remaining-context display; Context: 1000000
FrontierSWE v269.9%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-
Vibe Code Bench93.77%Vals AI, Vibe Code Bench v1.1 leaderboardchecked 2026-11-09Harness: OpenHands
FrontierCode 2.049.1%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more pageEffort: xhigh
CursorBench 4.060%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max; Harness: Cursor production harness
54.8%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: medium; Harness: Cursor production harness
53.2%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: low; Harness: Cursor production harness
56.9%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: high; Harness: Cursor production harness
56.3%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: xhigh; Harness: Cursor production harness
SWE-Bench Multimodal74.1%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
FrontierCode 2.0 (average cost per task)lower is better$2.48Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: medium; Harness: Claude Code
$44.7Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max; Harness: Claude Code
$6.72Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: high; Harness: Claude Code
$14.19Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: xhigh; Harness: Claude Code
$2.1Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: low; Harness: Claude Code

Science and Math

TestValuePrinted bySetup as printed
GPQA Diamond93.9%Epoch AI, Epoch AI Benchmarking Hubchecked 2026-11-10Effort: max
Terminal-Bench Science 0.174.6%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-
FrontierMath Tier 4 (v2)100%Epoch AI, Epoch AI Benchmarking Hubchecked 2026-11-10Effort: max
FrontierMath Tier 1-3 (v2)95.1%Epoch AI, Epoch AI Benchmarking Hubchecked 2026-11-10Effort: max

Computer Use

TestValuePrinted bySetup as printed
OSWorld 2.1 offline subset89.3%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-
OSWorld 2.1 binary54.6%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09-

Multimodal Understanding

TestValuePrinted bySetup as printed
Chartography (no tools)73.4%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-
Chartography (with tools)91.6%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max

General Knowledge and Reasoning

TestValuePrinted bySetup as printed
Humanity's Last Exam (no tools)73.2%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
Humanity's Last Exam (with tools)74.2%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-
SimpleQA Verified94%Epoch AI, Epoch AI Benchmarking Hubchecked 2026-11-10Effort: max
ARC-AGI-379.26%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: high; Harness: Standard
65.66%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max; Harness: Standard
ARC-AGI-3 (Provider Adapter harness)97.72%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: high; Harness: Provider Adapter
94.9%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max; Harness: Provider Adapter

Web Search and Research

TestValuePrinted bySetup as printed
WANDR (relaxed score)82.1%Anthropic, Claude Fable 5.2 and Mythos 5.2checked 2026-11-09; printed on 1 more page-

Multilingual

TestValuePrinted bySetup as printed
Global-MMLU96%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max
MILU94.1%Anthropic, System Card: Claude Fable 5.2 & Claude Mythos 5.2checked 2026-11-09Effort: max

Healthcare

Engineering and Design

Price

API list prices in US dollars per one million tokens, as printed on the developer's pricing page on the day checked. Cost per task is measured by Vals AI.

PriceValuePrinted bySetup as printed
Input$8Anthropic, Anthropic API pricingchecked 2026-11-09-
Cached input$0.2Anthropic, Anthropic API pricingchecked 2026-11-09-
Cache write$10Anthropic, Anthropic API pricingchecked 2026-11-09Cache: 5 minute cache
Cache write, 1 hour$16Anthropic, Anthropic API pricingchecked 2026-11-09-
Output$40Anthropic, Anthropic API pricingchecked 2026-11-09-
Input, batch$4Anthropic, Anthropic API pricingchecked 2026-11-09-
Output, batch$20Anthropic, Anthropic API pricingchecked 2026-11-09-
Input, fast$16Anthropic, Anthropic API pricingchecked 2026-11-09-
Output, fast$80Anthropic, Anthropic API pricingchecked 2026-11-09-
Cost per taskVals AI, USD, lower is better$22.13Vals AI, Vibe Code Bench v1.1 leaderboardchecked 2026-11-09Harness: OpenHands

Side by side

More from Anthropic

Every number here is read from the same catalog file as the benchmark table, data version 2026111201. Where one page printed the model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.