Inference
EN

Claude Fable 5 vs DeepSeek V4.1 Flash 1017

Change models
Claude Fable 5Anthropic
DeepSeek V4.1 Flash 1017DeepSeekFastest
IndexAll categories, out of 100.
74.0 ±2.7 best
68.5 ±2.6
GapIndex points below the top.
Top
−5.5
CoverageIndex weight that has results.
95%
95%
SpeedOutput tokens per second.
56/s
261/s best
EloLMArena rating and rank.
1492 best
1463
ReleasedMonth of release.
2026-06
2026-10
Input / 1MUS dollars per 1M input tokens.
$10.00
$0.022 best
Output / 1MUS dollars per 1M output tokens.
$50.00
$0.40 best
Batch output / 1MBatch price per 1M output tokens.
$25.00
$0.20 best
1K chats2K tokens in, 500 out, each.
$45.00
$0.244 best
1K documents20K tokens in, 1K out, each.
$250.00
$0.84 best
1K agent tasks200K tokens in, 20K out, each.
$3,000.00
$12.40 best
ContextMaximum tokens in one request.
1M
1.05M best
AgenticMulti-step tasks with tools.
71.4
72.2 best
CodingCode writing and repair.
74.7 best
69.8
ReasoningLogic problems and puzzles.
73.6 best
65.0
MultimodalTasks with images and text.
70.1 best
62.2
KnowledgeFacts and expert knowledge.
75.8 best
64.2
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
64.1
68.3 best
MathMath problems.
76.2 best
65.4

Benchmarks

52 shared
Only Claude Fable 5: 40Only DeepSeek V4.1 Flash 1017: 42
Claude Fable 5
DeepSeek V4.1 Flash 1017
Artificial Analysis Humanity's Last ExamKnowledge
55.5% best5 / 222
39.2%51 / 222
Artificial Analysis Long Context ReasoningReasoning
82.3%27 / 217
84.0% best8 / 217
GPQA diamond · Epoch AIKnowledge
85.9%69 / 217
89.8% best42 / 217
Critical Physics TasksReasoning
28.6% best14 / 214
14.3%52 / 214
OTIS Mock AIME 2024-2025Math
99.7% best14 / 200
98.3%28 / 200
Artificial Analysis Omniscience AccuracyKnowledge
65.4% best4 / 164
46.4%33 / 164
LegalBenchLegal
88.6% best1 / 149
83.3%56 / 149
GDPval-AA normalizedAgentic
55.5% best18 / 133
55.0%21 / 133
ARC-AGI-2 (semi-private)Reasoning
89.2% best11 / 131
72.9%25 / 131
ARC-AGI-1 (semi-private)Reasoning
98.5% best1 / 126
94.5%21 / 126
Vibe Code Bench v1.1Coding
90.4% best4 / 123
84.7%18 / 123
ARC-AGI-2 (public eval)Reasoning
96.4% best9 / 122
81.7%23 / 122
ARC-AGI-1 (public eval)Reasoning
97.6%17 / 116
98.0% best11 / 116
Artificial Analysis SciCodeCoding
61.0% best6 / 114
51.9%44 / 114
MedScribeHealthcare
88.5% best12 / 109
85.5%27 / 109
MedCodeHealthcare
56.1% best4 / 106
41.2%56 / 106
SimpleQA VerifiedKnowledge
70.7% best9 / 91
42.9%47 / 91
Chess PuzzlesReasoning
47.0%24 / 90
48.0% best19 / 90
FrontierMath-Tiers-1-3-v2-PrivateMath
87.0% best12 / 90
67.4%33 / 90
Legal Research BenchLegal
49.5% best9 / 90
41.3%30 / 90