Which is better, Claude Fable 5 or Gemini 3.5 Flash?

Across 48 shared benchmarks tracked on Vector Wire, Claude Fable 5 leads on 40 and Gemini 3.5 Flash leads on 8 — Claude Fable 5 leads on more. Strengths differ by axis; the table below shows every head-to-head number with provenance.

SEEDRefresh Models —Deals —Regulatory —Sources —

Claude Fable 5 vs Gemini 3.5 Flash

AnthropicvsGoogle48 shared benchmarks40–8 head-to-head

Benchmark	Claude Fable 5	Gemini 3.5 Flash
AA Agentic Index	56.6	70.4
AA Intelligence	62.1	52
AA-LCR	76.7	81
AA-Omniscience	43.3	21.2
arc_agi_1	90.5	92.5
ARC-AGI-2	76.8	72.1
arena_elo	1508	1477
arena_text_factuality	1486	1466
Artificial Analysis Coding Index	76.5	70.1
Blueprint-Bench 2	38.6	33.6
charxiv_rq	89.4	84.2
coding_arena_elo	1626	1490
critpt	28.6	13.1
DeepSWE	70	37
DeepSWE 1.1	70	37
Finance Agent v2	56.3	57.9
gdpval	61.9	57.8
GDPval-AA (Elo)	1932	1656
GDPval-AA v2	1760	1357
GPQA Diamond	92.6	92.2
HiL-Bench	56.3	27.7
HLE	64.5	42.7
IFBench	63.5	76.3
Legal Agent Benchmark	13.3	0.8
Legal Agent Benchmark (Harvey's Held-Out Set)	13.3	0.8
livebench	78.3	75
livebench_agentic_coding	62.2	49
livebench_coding	86	78.2
livebench_data_analysis	80.5	64.9
livebench_instruction_following	75.8	75.6
livebench_language	90.7	84.6
livebench_math	96	88.2
livebench_reasoning	89.7	82
MCP Atlas	84.7	83.6
MMMU-Pro	84.2	84.3
OmniScience Accuracy	65.3	51.4
OmniScience Non-Hallucination	36.4	38.2
OSWorld-Verified	85	78.4
scicode	60.2	53.1
simplebench	81.9	76.7
simpleqa_verified	68.3	68.4
SWE-bench Pro	80	55.1
TauBench V3 - Banking	38.1	32.2
Terminal-Bench 2.1	88	78.7
Terminal-Bench Hard	62.9	46.2
toolathlon	61.7	56.5
τ²-Bench Telecom (AA run)	98.5	95.6
τ³-Bench	26.8	25.4

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.