Anthropic•Rank #3NEW

Claude Fable 5.2

Proprietarygeneralreasoningcodingagent

Leaderboard Score

69.5

Scale: 0 - 100

Throughput Speed

85 tps

Context Window

API Pricing / 1M

.00

Parameter Scale

Proprietary

Monthly API Cost Calculator

Claude Fable 5.2 rates: $8/1M input • $40/1M output. Pick a workload preset or enter your own token volumes.

Monthly input (M tokens)Monthly output (M tokens)

Same workload on published-rate alternatives

GPT-5.2 Pro (OpenAI)

GPT-6 Astra (OpenAI)

Open the comparison engine →

Estimates use listed API rates at time of writing; providers may differ in caching discounts, rate limits, and batching pricing.

Sourced AI Radar & Benchmarks•9 Evaluated Metrics

Capability Radar & Benchmark Scores

Multi-Axis Capability Radar

Comparing Claude Fable 5.2 against Frontier Top 10 & Open-Source average

0 - 100 Scale

Claude Fable 5.2

Frontier Top 10 Avg

Open-Source Avg

Performance Vectors

Where Claude Fable 5.2 Leads

Coding & SWETop 10: 79.2% • OS: 51.6%

82.5%+3.3% vs Top 10

Reasoning & MathTop 10: 94.6% • OS: 54.3%

95.6%+1.0% vs Top 10

Agent & Tool AutomationTop 10: 80.3% • OS: 37.4%

89.3%+9.0% vs Top 10

Terminal & CLI ExecutionTop 10: 63.5% • OS: 45.8%

68.0%+4.5% vs Top 10

MMLU & Knowledge QATop 10: 50% • OS: 86.7%

86.9%+36.9% vs Top 10

Cyber & Security ExploitsTop 10: 87.3% • OS: 81.2%

66.0%-21.3% vs Top 10

All Verified Benchmark Evaluations

Direct Pass@1 and resolution rates across standardized evaluations.

OverallCodingReasoningAgenticTerminal

Featured Compute Platform

Run Benchmarks on Dedicated AI Clusters

Zero-cold-start inference endpoints with dedicated GPU instances.

Compare With Alternatives →

Executive Analysis & Verdict

Claude Fable 5.2 by Anthropic is a next-generation frontier agent model ranking #3 globally with a 69.5 overall score, 69.9% FrontierSWE v2, 89.3% OSWorld 2.1, and 74.2% Humanity's Last Exam.

Key Strengths & Highlights

•Global leadership on FrontierSWE v2 (69.9%) and FrontierCode 2.0 (49.1% Xhigh effort)
•Exceptional computer use accuracy with 89.3% on OSWorld 2.1 (offline subset)
•20% price reduction to $8 / $40 per 1M tokens with optional 2.5x Fast Mode

Limitations & Considerations

•Fast Mode carries premium pricing ($16 / $80 per 1M tokens)

Recommended Workflows

Autonomous agentic coding, multi-step computer use, long-horizon scientific research, and complex enterprise automation

Architectural Overview & Capability Profile

Claude Fable 5.2 is Anthropic’s flagship frontier agent foundation model, co-released alongside Claude Mythos 5.2. Engineered specifically to excel across autonomous agent loops, continuous computer use, and complex software engineering, Claude Fable 5.2 achieves an overall RankLLMs rating of 69.5 / 100, placing it solidly at #3 globally directly alongside Gemini 4 Carbon and Claude Opus 5.5.

Introduced with a 20% price reduction over Fable 5.1 to $8.00 per 1M input tokens and $40.00 per 1M output tokens, Fable 5.2 provides a 1,000,000-token context window, native multimodal processing, and an optional Fast Mode that accelerates token generation throughput by up to 2.5x.

Official Anthropic Benchmark Suite

Across rigorous evaluations published by Anthropic comparing Fable 5.2 directly against Fable 5.1, Opus 5.5, GPT-6 Astra, and Gemini 4 Carbon, Claude Fable 5.2 secures decisive wins in autonomous execution:

Agentic Software Engineering & Coding:
- FrontierSWE v2: Scores 69.9%, leading GPT-6 Astra (65.5%), Opus 5.5 (62.3%), Gemini 4 Carbon (61.2%), and Fable 5.1 (56.3%).
- FrontierCode 2.0: Reaches 49.1% under Xhigh effort, outperforming Opus 5.5 (45.5%), GPT-6 Astra (42.8%), and Gemini 4 Carbon (41.1%).
Computer Use & Workplace Tool Execution:
- OSWorld 2.1 (Offline subset): Sets a new benchmark high of 89.3%, outpacing Opus 5.5 (87.2%), Fable 5.1 (83.5%), Gemini 4 Carbon (81.5%), and GPT-6 Astra (79.5%).
- AutomationBench: Scores 51.4%, nearly doubling Fable 5.1 (31.4%) and beating GPT-6 Astra (41.4%) and Opus 5.5 (42.5%).
- Remote Labor Index (RLI): Achieves 41.3% (High effort), more than double Fable 5.1 (17.9%) and Opus 5.5 (21.3%).
- RoboDojo Sim: Captures 64.9%, significantly advancing physical simulation and robotics planning over previous baselines.
Multidisciplinary Reasoning & Scientific Research:
- Humanity’s Last Exam (with tools): Achieves 74.2%, leading Opus 5.5 (67.7%), Fable 5.1 (65.6%), and GPT-6 Astra (57.2%).
- Terminal-Bench-Science 0.1: Reaches 74.6%, leading GPT-6 Astra (68.1%) and Gemini 4 Carbon (64.8%).
- WANDR (Agentic Search): Reaches 82.1% relaxed score, surpassing Opus 5.5 (72.3%) and Fable 5.1 (68.7%).
- Chartography (Visual Chart Understanding): Scores 73.4% without tools, outpacing Gemini 4 Carbon (72.5%) and GPT-6 Astra (71.0%).

Deployment & Fast Mode Economics

Claude Fable 5.2 is available across the Anthropic API, Claude Code, and major cloud providers:

Standard Tier: $8.00 / 1M input and $40.00 / 1M output with $0.20 / 1M prompt caching read.
Fast Mode: Available for latency-critical agent loops at $16.00 / 1M input and $80.00 / 1M output, delivering up to 180 tokens/sec.

Frequently Asked Questions About Claude Fable 5.2

What is Claude Fable 5.2 and who developed it?↓

Claude Fable 5.2 is a proprietary large language model developed by Anthropic. It ranks #3 on the RankLLMs composite leaderboard, ahead of 98% of the 104 models we track, with an overall score of 69.5 / 100.

Can Claude Fable 5.2 handle very long documents and codebases?↓

Claude Fable 5.2 supports a 1.0M-token input context window and generates about 85 tps tokens per second in our reference measurements and which is enough to hold entire repositories, long transcripts, or multi-document research sets in a single session..

Why does Claude Fable 5.2 cost more than most alternatives?↓

Claude Fable 5.2 sits in the premium tier at .00 per 1M tokens - cheaper than only 95% of priced models we track. That pricing buys frontier-tier benchmark results; the cost-per-solved-task math in our best-coding-LLM guide shows when it pays off.

What is Claude Fable 5.2 best at?↓

Claude Fable 5.2 is strongest in Global leadership on FrontierSWE v2 (69.9%) and FrontierCode 2.0 (49.1% Xhigh effort); Exceptional computer use accuracy with 89.3% on OSWorld 2.1 (offline subset); 20% price reduction to $8 / $40 per 1M tokens with optional 2.5x Fast Mode. Full pillar-by-pillar numbers are in the benchmark tables above.

What should you watch out for with Claude Fable 5.2?↓

Known trade-offs include: Fast Mode carries premium pricing ($16 / $80 per 1M tokens). Weigh these against the strengths above for your specific workload.