Claude Fable 5.1 vs GPT-6 Astra

GPT-6 Astra leads on the Visual Intelligence Index by 6.0 points, and Claude Fable 5.1 scores higher on 1 of the 7 benchmarks where the two differ.

Model

Compared with

At a glance

Measure	Claude Fable 5.1	GPT-6 Astra
Visual Intelligence Index	76.2%	82.2%
Rank	#3 of 9	#1 of 9
Cost per taskMean over the benchmarks, at list price.	$0.517	$0.864
Latency per taskEach benchmark's median, averaged.	9.0s	9.6s
Tokens per task	51,151	86,253
Input / 1M tokens	$10.00	$10.00
Output / 1M tokens	$50.00	$50.00
Context window	1M	1.05M
Availability	Proprietary	Proprietary

Each capability index is a weighted mean of the benchmarks that contribute to the capability.

Capability	Claude Fable 5.1	GPT-6 Astra	Lead
Perception	75.3%	80.8%	−5.5
Spatial Reasoning	77.0%	84.6%	−7.5
Temporal Reasoning	76.4%	83.0%	−6.6
Action / Event Understanding	74.0%	79.2%	−5.3
Causal Reasoning	78.7%	83.8%	−5.0
Information Retrieval	74.7%	80.7%	−6.0

Benchmark	Claude Fable 5.1	GPT-6 Astra	Lead
Video-MME v2	66.2%	73.4%	−7.2
LVBench	75.0%	81.6%	−6.6
Perception Test	77.3%	92.5%	−15.2
NExT-QA	86.8%	90.7%	−3.9
Q-Bench Video	71.0%	68.3%	+2.7
EgoSchema	80.4%	83.4%	−3.0
UCF101-AD	60.6%	61.4%	−0.8

3.0/ 5

Reachable1.0: There is an endpoint you can call without hosting anything.
First-party API — supported
Third-party API — supported
Portable0.0: You can run it yourself, and are not tied to one vendor.
Open weights — not supported
Self-hostable — not supported
Multimodal input0.0: It takes the footage directly, rather than frames you extracted.
Video ingestion — not supported
Audio — not supported
Programmable1.0: Output you can parse, and tools it can call on its own.
Structured output — supported
Tool calling — supported
Operable1.0: Usable interactively and in bulk, not only one call at a time.
Streaming — supported
Batch API — supported

3.5/ 5

Reachable1.0: There is an endpoint you can call without hosting anything.
First-party API — supported
Third-party API — supported
Portable0.0: You can run it yourself, and are not tied to one vendor.
Open weights — not supported
Self-hostable — not supported
Multimodal input0.5: It takes the footage directly, rather than frames you extracted.
Video ingestion — not supported
Audio — supported
Programmable1.0: Output you can parse, and tools it can call on its own.
Structured output — supported
Tool calling — supported
Operable1.0: Usable interactively and in bulk, not only one call at a time.
Streaming — supported
Batch API — supported