Anthropic doubles Opus 5.5 fast mode pricing to $8/$40 per million tokens
Anthropic’s Opus 5.5 fast mode now costs $8 per million input tokens and $40 per million output tokens, double the standard rates of $4/$20. The feature accelerates output token generation by up to 2.5x but does not reduce the time to first token. Anthropic’s official documentation confirms the same model is used, with only inference speed altered.
Key points
- Opus 5.5 fast mode doubles pricing to $8/$40 per million tokens, matching 2.5x output speed but no TTFT change
- Savings per hour vary from $5–$29 depending on output length and assumed token speeds (40–80/sec)
- Anthropic’s documentation excludes Batch API support and warns of rate limits; switching back resets prompt cache
The cost-effectiveness of fast mode depends on output length. For a 3,000-token response, the price increase is $0.076 per run, saving about 45 seconds at 40 tokens/second. However, for shorter outputs (e.g., 300 tokens), the savings drop to 2–4 seconds, making the $5–$29/hour estimate for saved time less compelling. The author notes that standard output speeds are unlisted, requiring assumptions (40, 60, or 80 tokens/sec) to calculate savings. Background tasks or long outputs may justify the cost, while short, real-time tasks likely won’t.
The article also compares Opus 5.5 fast mode to the cheaper Sonnet 5.5 ($2/$10), cautioning users to weigh model performance against speed needs before opting for the premium tier.
Model page: Claude Sonnet 5.5 →
Claude Opus 5.5 fast mode costs double; the price to buy 1 hour of waiting time is about $5 to $29
note.com · 29 October 2026
Loading the full article…
This text was published by note.com and written by AI Scout JP. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Google announces Gemini 4 Carbon and Genie 4 models with staged rollout for cyber defenders · 7 src
- LiquidAI releases browser demo for LFM3-Omni voice AI · 1 src
- Datalab releases Surya 2.5 OCR model with 85.9% accuracy on OmniParseBench · 1 src
- NVIDIA releases open-weight Nemotron 3.5 Transcribe model and Granary v2 dataset · 13 src
- Mistral releases Mistral Large 4 (Le Chonk) 1-trillion-parameter multimodal model on Hugging Face · 11 src
Comments
via GitHub Discussions