Coin Bureau reports that Claude Opus 5.5 launched with pricing of $4 per million input tokens and $20 per million output tokens. It also relays Anthropic’s claim that the model matches Fable 5.1 on most tasks and leads Anthropic’s lineup on coding and knowledge-work benchmarks. In a two-hour SEO test, a user says Opus reviewed 347 ranking pages across 40 keywords and found 131 content angles those pages did not cover, at a cost of $6.35. The same user says Fable handled 16 of the same keywords in the same time for $5.53—an account of one test, rather than a general performance comparison.
Featured Source
borja · Watch on X🚨 Claude Opus 5.5 just launched
i tested it for unique SEO content!!!
a.k.a. information gain
i spent the last 2 hours since Claude Opus 5.5 launched testing this
- it read the google top 10 for 40 of my keywords (347 ranking pages)
- mapped what every page already says
- found 131 angles nobody covers yet and scraped raw data to come up with unique insights
- 3.3 angles per keyword.
total cost $6.35.
Claude Fable 5.1, same keywords, same clock, got through 16 and spent $5.53. same 3.4 angles per keyword, so no smarter, just slower.
it's 60% cheaper per token. the full Fable pass would have run $16.38. and Opus broke 1 of my 8 writing rules in 131 drafts. Fable broke 2 in 37.
Opus sticks to writing rules better
information gain is the whole game now: say something the top 10 results don't.
Why these 2 models? Because they're the best. And I want the best for something as delicate as Information Gain
Example: This same post is an experiment I ran and the findings are considered information gain
Now powering content research at http://distribb.io
p.s. thanx for the credit reset anthropic, i was crying this morning...
Coin Bureau reports that Claude Opus 5.5 launched with pricing of $4 per million input tokens and $20 per million output tokens. It also relays Anthropic’s claim that the model matches Fable 5.1 on most tasks and leads Anthropic’s lineup on coding and knowledge-work benchmarks. In a two-hour SEO test, a user says Opus reviewed 347 ranking pages across 40 keywords and found 131 content angles those pages did not cover, at a cost of $6.35. The same user says Fable handled 16 of the same keywords in the same time for $5.53—an account of one test, rather than a general performance comparison.
Featured Source
borja · Watch on X🚨 Claude Opus 5.5 just launched
i tested it for unique SEO content!!!
a.k.a. information gain
i spent the last 2 hours since Claude Opus 5.5 launched testing this
- it read the google top 10 for 40 of my keywords (347 ranking pages)
- mapped what every page already says
- found 131 angles nobody covers yet and scraped raw data to come up with unique insights
- 3.3 angles per keyword.
total cost $6.35.
Claude Fable 5.1, same keywords, same clock, got through 16 and spent $5.53. same 3.4 angles per keyword, so no smarter, just slower.
it's 60% cheaper per token. the full Fable pass would have run $16.38. and Opus broke 1 of my 8 writing rules in 131 drafts. Fable broke 2 in 37.
Opus sticks to writing rules better
information gain is the whole game now: say something the top 10 results don't.
Why these 2 models? Because they're the best. And I want the best for something as delicate as Information Gain
Example: This same post is an experiment I ran and the findings are considered information gain
Now powering content research at http://distribb.io
p.s. thanx for the credit reset anthropic, i was crying this morning...
AI