Compare models for your API workload

Up to four models side by side: what each costs you a month, where and how you call it, which features it supports, and what its card says.

Your workloadRAG over docs · 12K in · 600 out · 50K requests/mo · 30% cached

Models · 3 of 4 pickedShow

Anthropic

OpenAI

Google

xAI

Meta

Alibaba

Mistral AI

DeepSeek

ByteDance

	Claude Fable 5.1Anthropic	GPT-6 AstraOpenAI	Gemini 3.1 ProGoogle
Your cost (list prices, your workload)
Per month	$5,7454.6×	$5,8804.8×	$1,236cheapest
Per 1K requests	$115	$118	$24.72
Call it
Maker's own API	`claude-fable-5-1` Claude API	`gpt-6-astra` OpenAI API	`gemini-3.1-pro-preview` Gemini API
Amazon Bedrock	`anthropic.claude-fable-5-1`	—	—
Google Vertex AI	`claude-fable-5-1`	—	`gemini-3.1-pro-preview`
Microsoft Foundry / Azure	`claude-fable-5-1`	—	—
Gotchas	Thinking can't be turned off No forced tool_choice No prefill or sampling params	Reasoning can't be turned off Tool calling needs the Responses API No sampling params while reasoning	Still a preview model Higher price above 200K tokens Separate custom-tools endpoint
Supported features (from each maker's docs)Differences only
Tool calling	Yes	Yes	Yes
Structured outputs	Yes	Yes	Yes
Image input	Yes	Yes	Yes
PDF input	Yes	Not documented	Yes
Prompt caching	Yes	Yes	Yes
Batch API	Yes	Yes	Yes
Streaming	Not documented	Yes	Not documented
Reasoning control	Yes	Yes	Yes
Web search	Yes	Yes	Not documented
Fine-tuning	Not documented	Not documented	Not documented
Specs
Input / 1M	$10.00	$10.00	$2.00
Output / 1M	$50.00	$50.00	$12.00
Cache read / 1M	$0.25	$1.00	$0.20
Context	1M	1.05M	1M
Max output	128K	128K	64K
Knows up to	June 2026	April 2026	January 2025
Reasoning	Adaptive (always on)	Reasoning, always on (effort low → max)	Dynamic, thinking_level low / medium / high
Default effort	high	—	high
Inputs	text, image, pdf	text, image	text, image, audio, video, pdf
Weights	Closed	Closed	Closed
Released	1 Sep 2026	3 Sep 2026	19 Feb 2026
API model ID	claude-fable-5-1	gpt-6-astra	gemini-3.1-pro-preview
From each card
Built for	Coding, knowledge work and problem-solving, including novel mathematical and scientific reasoning Largest gains over Fable 5 in terminal-based scientific and engineering work, computer use, and long-horizon agentic and professional work Matches or exceeds Fable 5 at roughly half the cost per task on agentic coding benchmarks	Complex reasoning and coding Computer use and browsing Research and document creation	Agentic performance Advanced coding Long context and multimodal understanding
Watch out for	Prompt injection gets through on fallback turns Overstates what the user authorized Guesses credentials in computer use	Oversteps permissions in agentic coding Out-of-scope attacks in hard cyber tasks Harder to monitor through its reasoning	Cyber capability at alert threshold Hallucinations Old knowledge cutoff
Source	System Card: Claude Fable 5.1 & Claude Mythos 5.1 ↗	GPT-6 Astra System Card ↗	Gemini 3.1 Pro Model Card ↗
Benchmarks reported on more than one card (self-reported, settings differ)
Terminal-Bench 4.0	55.8	57.9best	—
Terminal-Bench-Science 0.1	52.6	64.6best	—
Humanity's Last Exam (no tools)	60.9best	—	44.4
Humanity's Last Exam (with tools)	65.0best	57.2	—
HealthBench Professionalscore	62.1	64.7best	—

Benchmark numbers come from each maker's own card. Makers run them with different settings and harnesses, so treat cross-maker differences as rough.