model comparison

vs

Claude Fable 5.2 vs GPT-6 Astra

Anthropic against OpenAI, compared on listed price, real workload cost, context and capabilities.

the short answer

  • on Humanity's Last Exam (with tools), as each maker reports it, Claude Fable 5.2 scores 74.2% and GPT-6 Astra 57.2%.
  • Claude Fable 5.2 lists lower input and output prices.
  • Claude Fable 5.2 reads cached input for less ($0.2/M against GPT-6 Astra's $1/M), which decides the bill for agents that resend long context.

side-by-side specs

specClaude Fable 5.2GPT-6 Astra
providerAnthropicOpenAI
input price$8/M$10/M
output price$40/M$50/M
cached input$0.2/M$1/M
blended price (3:1)$16/M$20/M
context window1M tokens1M tokens
max output128k tokens128k tokens
inputtext, imagestext, images
structured outputyesyes
weightsclosedclosed
releasedNovember 9, 2026September 2026
Humanity's Last Exam (with tools) (maker-reported)74.2%57.2%

same work, both models

workloadClaude Fable 5.2GPT-6 Astra
one chat reply3,000 tokens in, 500 outGPT-6 Astra costs 1.3× as much as Claude Fable 5.2$0.04$0.06
summarise a 100-page report50,000 tokens in, 1,500 out, one requestGPT-6 Astra costs 1.2× as much as Claude Fable 5.2$0.46$0.58
1,000 classification calls500 tokens in, 20 out, per callGPT-6 Astra costs 1.3× as much as Claude Fable 5.2$4.80$6.00
coding-agent session2M tokens in across ~60 turns, 60,000 out, no cache$18.40$23.00
same session, cached85% of input read from the prompt cache, the rest written to itGPT-6 Astra costs 1.5× as much as Claude Fable 5.2$5.74$8.45

at listed rates. token counts are typical sizes, not measurements.

Claude Fable 5.2 pricing notes

  • cache hits bill at $0.20 per 1M tokens (0.025x input); cache writes at $10 (5-minute tier) or $16 (1-hour tier).
  • the full 1M-token window bills at the standard rate, with no long-context surcharge.
  • adaptive thinking is always on and cannot be disabled; the effort setting (default high) steers its depth.

GPT-6 Astra pricing notes

  • requests over 272K input tokens bill in full at 2x input and cache rates and 1.5x output; the listed price is the ≤272K rate.
  • cache writes bill at $12.50 per 1M tokens (30-minute tier); cache hits at $1.00.

when to pick which

Claude Fable 5.2

Anthropic's Claude model for demanding reasoning and long-horizon agentic work, released as the successor to Fable 5.1. It costs less than 5.1 on input, output and cache, at the same context window; Anthropic points most workloads to the cheaper Opus 5.5.

pick it if

  • Opus 5.5 at higher effort still falls short on your evals
  • you run Fable 5.1 today and want lower rates on the same window
  • you run long agent sessions; cache hits bill at 2.5% of input

skip it if

  • Opus 5.5 covers your workload at half the per-token price
  • your code forces tool use, prefills replies, sets temperature or edits earlier turns; all return errors

GPT-6 Astra

OpenAI's top GPT-6 model, which OpenAI calls its most capable for the most demanding work. it takes text and images and costs five times GPT-6.1 Sol's per-token rate.

pick it if

  • you want OpenAI's top tier for complex reasoning, coding or computer use
  • you resend long prompts; cache hits bill at 10% of input
  • you need replies of up to 128K output tokens

skip it if

  • GPT-6.1 Sol covers your workload at one-fifth the per-token price
  • your prompts routinely pass 272K tokens; the whole request then bills higher

faq

which is better, Claude Fable 5.2 or GPT-6 Astra?

on Humanity's Last Exam (with tools), as each maker reports it, Claude Fable 5.2 scores 74.2% and GPT-6 Astra 57.2%. these are each maker's own numbers, not a controlled head-to-head; the spec table lists every benchmark both publish.

which is cheaper, Claude Fable 5.2 or GPT-6 Astra?

Claude Fable 5.2 lists lower input and output prices. on a coding-agent session (2M tokens in, 60,000 out) Claude Fable 5.2 costs $18.40 and GPT-6 Astra $23.00.

can i switch between Claude Fable 5.2 and GPT-6 Astra mid-task?

yes. in an octomind session, /model octohub:claude-fable-5-2 and /model octohub:gpt-6-astra switch models without losing the conversation, so you can draft on one and review on the other.

try either model

switch between models inside the same octomind session.

open the panel all models

prices from the catalog of 2026-11-10