← The Polylog AI Intelligence Brief
Morning Edition · Thursday, September 24, 2026Published at 2:18 AM EDT · New York
Anthropic Prices Claude Opus 5.5 at $4 and $20 per Million Tokens, Undercutting Its Own Flagship
Anthropic claims Opus 5.5 matches the quality of Fable 5.1 on most work at roughly 40 percent lower total cost than Opus 5, with about half of that saving coming from the model using fewer tokens per task rather than from the lower list price.
Anthropic's Claude Opus 5.5 is listed at $4 per million input tokens and $20 per million output tokens, about 20 percent below Opus 5 on list price alone. The company's larger claim, that typical workloads cost roughly 40 percent less overall, rests on the model completing tasks using fewer tokens because it reasons more efficiently. Those are two separate figures, and only the list price is a fixed, contractual number.
Compared with Anthropic's own top model, the price gap is wider still. VentureBeat reported that Opus 5.5 beats Claude Fable 5.1 on key agentic benchmarks at roughly 60 percent lower application programming interface (API) price. The benchmark suite belongs to Anthropic, covering agentic coding, computer use, and knowledge work, and no independent group has reproduced the results yet. On two evaluations where Anthropic does not lead, Terminal-Bench-Science and AutomationBench, OpenAI's GPT-6 Astra still holds the top position by Anthropic's own accounting, a detail that tempers the marketing claims.
A separate, practical demonstration circulated online. A three-minute video tracing the field's history from the 2017 paper "Attention Is All You Need" to current systems was rendered entirely by code that Opus 5.5 generated, with no stock footage and no image-generation models in the pipeline. That is a long-horizon code-generation task, precisely the kind of workload the new pricing is meant to win.
The pattern holds across vendors. The most advanced available capability is now being delivered at falling cost per token, and labs are choosing to cut prices on their second-tier model rather than protect profit margin on their top model.
What this means
Anthropic is deliberately taking revenue away from its own Fable 5.1 model to hold market share in agentic coding, and that move compresses profit margins across every vendor that sells tokens for long-running agent tasks. Application companies that resell model output, from coding tools to legal-drafting software, benefit immediately because their costs fall without any change to their own product. The companies most exposed are vendors whose pricing assumes customers cannot easily switch models, because a 40 percent saving on a workload is large enough to justify rebuilding an agent system around a different provider.
Part of a tracked trend
Frontier Model Price War
Frontier API vendors increasingly compete on strategic price-cutting rather than pure capability, repeatedly launching or repricing models below prevailing rates to grab share and compressing industry inference margins; expect recurring below-rival pricing moves.
Townsquare
Start a discussion in Townsquare.
More from this edition
- Anthropic Says 950 Claude Agents Found a New CRISPR-Like Enzyme System in 21 Hours
- ChatGPT Voice Gains Access to Email, Calendar and Slack, and the Ability to Act on Them
- Meta Prices New VR Glasses at $1,299 and Puts Its Muse Agent on a Keychain
- Altman and Amodei Ask the UN Security Council for Binding Standards on Frontier AI
- Qualcomm Commits to Linux on Snapdragon X2 Laptops, With Debian First and Ubuntu Certification in 2027
- OpenAI Opens Its Daybreak Cyber Program to Ukraine's Government
- Google Adds Encrypted Server-Side Memory to Private AI Compute, With Keys Held on the User's Device
Comments
0No comments yet.