Anthropicproprietary
Released: 2026-11-09Artificial AnalysisClaude Fable 5.2
Claude Fable 5.2 is Anthropic's most capable model for the most demanding technical work, with gains over Fable 5.1 in coding, reasoning, and vision. In prerelease testing it completed otherwise-infeasible...
Intelligence Index
64.6/ 100
Artificial Analysis Intelligence Index (independent composite)Throughput Speed
102tok/sec
Output streaming throughputTime to First Answer
262s
Time to first answer token, including any thinkingCost to Run the Index
26.54USD
Artificial Analysis cost of the full Index runMeasured results · Artificial Analysis
Independent evaluations of Claude Fable 5.2 (Max, Default Fallback configuration). Values are reported as published; nothing here is estimated or derived. Snapshot 2026-11-15.
- Intelligence Index
- 64.6
- Cost to run the Index
- $26.54
- Time per task
- 44.5 min
- Omniscience index
- 66.8
- Reasoning model
- Yes
- Cached-input price
- $0.2 / 1M
Reasoning & knowledge
- AA-Omniscience Accuracy79.1%
- CritPt35.4%
- CritPt v2 (Hard)76.5%
- HLE-Diamond80.2%
- Humanity's Last Exam74.3%
Coding & terminal agents
- FrontierCode 1.156.2%
- FrontierCode-AA 2.048.8%
- SciCode67.7%
- Terminal-Bench 4.059.6%
- Terminal-Bench-Science 0.173.3%
Agentic work
- AA-Briefcase v1.1 (Elo)1785 Elo
- AutomationBench-AA76.4%
- GDPval-AA (normalized Elo)1831 Elo
- Harvey LAB-AA7.2%
- Multimodal Knowledge Agent (MKA)56.6%
Long context & documents
Multimodal & science
- AA-3D (Elo)1271 Elo
- VIALS-AA67.1%
Health & safety
Reasoning-effort variants evaluated
| Variant | Intelligence Index | Cost to run Index | Output speed (tok/s) |
|---|---|---|---|
| Max, Default Fallback | 64.6 | $26.54 | 102.4 |
| Xhigh, Default Fallback | 62.2 | $14.83 | 98.8 |
| High, Default Fallback | 61.8 | $9.08 | 105.1 |
| Medium, Default Fallback | 56.4 | $3.80 | 97.5 |
| Low, Default Fallback | 53.9 | $3.10 | 106.1 |
Headline Evaluation Breakdown
Six independently measured evaluations (Artificial Analysis)Reasoning · CritPt v2 (Hard)76.5%
Coding & agents · Terminal-Bench 4.059.6%
Long context · AA-LCR83.3%
Knowledge · Humanity's Last Exam74.3%
Sourced Benchmark Evaluations (20)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| AA-Omniscience (Non-Hallucination) | Active | 40.7% | 93 | 2026-11-15 | independent | Source ↗ |
| VIALS-AA | Active | 67.1% | 93 | 2026-11-15 | independent | Source ↗ |
| Harvey LAB-AA | Active | 7.2% | 93 | 2026-11-15 | independent | Source ↗ |
| AutomationBench-AA | Active | 76.4% | 93 | 2026-11-15 | independent | Source ↗ |
| Multimodal Knowledge Agent (MKA) | Active | 56.6% | 93 | 2026-11-15 | independent | Source ↗ |
| MLCR-AA (Medical Long Context Reasoning) | Active | 43.9% | 93 | 2026-11-15 | independent | Source ↗ |
| GDP.pdf | Active | 24.2% | 93 | 2026-11-15 | independent | Source ↗ |
| FrontierCode-AA 2.0 | Active | 48.8% | 93 | 2026-11-15 | independent | Source ↗ |
| HealthBench Professional | Active | 67.7% | 93 | 2026-11-15 | independent | Source ↗ |
| HealthBench | Active | 59.6% | 93 | 2026-11-15 | independent | Source ↗ |
| FrontierCode | Active | 56.2% | 98 | 2026-11-15 | independent | Source ↗ |
| AA-Omniscience (Accuracy) | Active | 79.1% | 93 | 2026-11-15 | independent | Source ↗ |
| HLE-Diamond | Active | 80.2% | 93 | 2026-11-15 | independent | Source ↗ |
| CritPt v2 (Hard) | Active | 76.5% | 93 | 2026-11-15 | independent | Source ↗ |
| CritPt | Active | 35.4% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-LCR (Long Context Reasoning) | Active | 83.3% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench-Science 0.1 | Active | 73.3% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench 4.0 | Active | 59.6% | 93 | 2026-11-15 | independent | Source ↗ |
| SciCode | Active | 67.7% | 93 | 2026-11-15 | independent | Source ↗ |
| Humanity's Last Exam | Active | 74.3% | 93 | 2026-11-15 | independent | Source ↗ |
Anthropic Python SDKAPI Ready
import anthropic
client = anthropic.Anthropic()
message = client.messages.create(
model="anthropic-claude-fable-5-2",
max_tokens=2048,
temperature=0,
messages=[{"role": "user", "content": "Execute SWE-bench verified repository fix"}]
)
print(message.content[0].text)Evaluation & Sourcing Notes
Scores listed for Claude Fable 5.2 represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.