OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?
Comments on “OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Between June and late September 2026, OpenAI and Anthropic cut effective prices sharply: GPT-5.6 Luna fell 80%, Opus 5.5 costs 40-60% less than Opus 5, and GPT-6.1 Sol is about 75% cheaper than the prior Sol. Anthropic says Opus 5.5 needs less compute to serve. The author attributes the cuts mainly to open-weight competition, with 29 providers hosting DeepSeek V4.1 Flash on OpenRouter. Several figures, including Vercel's cost index and the benchmark scores, were not verified.
Why it ranked
No. 18 of 62 with a score of 0.65. The panel rated it highest on new information (0.83) and audience breadth (0.81), lowest on substantiated (0.54).
Every answer behind this is in the table under Score.
Score
0.65total under rubric v2.2
- Significance0.67
- Reader interest0.77
- Audience breadth0.81
- Practical relevance0.70
- Substantiated0.54
- Technical depth0.54
- New information0.83
- Counts against: Promotional0.07
- Counts against: Overclaim0.45
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| Significance | 0.68 | 0.60 | 0.72 | 0.67 | 0.12 | +0.24 | +0.160 |
| Reader interest | 0.55 | 0.85 | 0.90 | 0.77 | 0.35 | +0.26 | +0.199 |
| Audience breadth | 0.82 | 0.80 | 0.80 | 0.81 | 0.02 | +0.13 | +0.105 |
| Practical relevance | 0.71 | 0.70 | 0.70 | 0.70 | 0.01 | +0.22 | +0.155 |
| Substantiated | 0.49 | 0.50 | 0.62 | 0.54 | 0.13 | +0.05 | +0.027 |
| Technical depth | 0.65 | 0.40 | 0.58 | 0.54 | 0.24 | +0.05 | +0.027 |
| New information | 0.93 | 0.60 | 0.96 | 0.83 | 0.36 | +0.05 | +0.042 |
| Promotional | 0.13 | 0.05 | 0.03 | 0.07 | 0.10 | −0.19 | −0.013 |
| Overclaim | 0.41 | 0.35 | 0.58 | 0.45 | 0.23 | −0.11 | −0.049 |
| Sum of contributions | +0.652 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.652 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-10-26/18/"><img src="https://rubric.hrih.si/badge/2026-10-26/18.svg" alt="Ranked #18 on Rubric, 26 Oct 2026" width="245" height="28"></a>Claim check
Claim check: 3 matches source · 5 couldn’t check. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
Couldn’t checkNumber
OpenAI cut the price of GPT-5.6 Luna by 80% to $0.20 / $1.20 / $0.02 on July 30, 2026.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “OpenAI announces that they cut the price of GPT-5.6 Luna by 80% (???) to $0.20 / $1.20 / $0.02”
The source says No passage could be compared.
Source cited by the article: openai.com/index/advancing-the-price-performance-frontier-with-gpt-5-6
Matches sourceNumber
Anthropic announced Fable 5.1 at $10 / $50 / $0.25, with strictly better benchmark performance than Fable 5 in all areas and a 75% cut in cache read cost.
The cited source says the same thing.
The article says “Anthropic announced Fable 5.1 at a price of $10 / $50 / $0.25, which, in addition to having strictly better benchmark performance in all areas than Fable 5”
The source says “Fable 5.1’s pricing is otherwise the same as Fable 5’s: $10 per million input tokens and $50 per million output tokens”
Source cited by the article: anthropic.com/claude-fable-and-mythos-5-1
Matches sourceQuote
Anthropic stated that Opus 5.5 requires less compute to serve than Opus 5 and its pricing reflects that.
The cited source says the same thing.
The article says “Opus 5.5 requires less compute to serve than Opus 5, and its pricing reflects that”
The source says “Opus 5.5 requires less compute to serve than Opus 5, and its pricing reflects that”
Source cited by the article: anthropic.com/claude-opus-5-5
Matches sourceNumber
Claude Opus 5.5 is priced at $4 / $20 / $0.20, a total cost reduction of 40-60% versus Opus 5, for a model better than Fable 5.1.
The cited source says the same thing.
The article says “Its price of $4 / $20 / $0.20 is much lower in total than Opus 5’s due to a 20% input/output cost cut and a larger cache read cut”
The source says “Input and output tokens are $4 and $20 per million, 20% less than Opus 5. Cache reads (which make up the majority of agentic and coding work costs) are $0.20…”
Source cited by the article: anthropic.com/claude-opus-5-5
Couldn’t checkNumber
Vercel's AI gateway data shows cost-per-token falling to 43.3% of the April 2026 cost.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “Data from Vercel’s AI model gateway supporting the decreasing cost-per-token trend to 43.3% of the April 2026 cost”
The source says No passage could be compared.
Source cited by the article: vercel.com/blog/ai-gateway-production-index-october-2026
Couldn’t checkBenchmark result
DeepSeek Flash V4.1 1017 scores 45 on the Artificial Analysis index versus 24 for DeepSeek V4 Flash.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “performs significantly better than DeepSeek Flash V4 benchmark-wise (45 on the Artificial Analysis index vs”
The source says No passage could be compared.
Source cited by the article: artificialanalysis.ai/models/deepseek-v4-1-flash-1017
Couldn’t checkBenchmark result
Kimi K3.1 Pro scores 55 on Artificial Analysis versus 58 for Opus 5.5, at half the input/output price and the same cache read cost.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “Kimi K3.1 Pro which has comparable performance to Opus 5.5 (Kimi at 55 vs. Opus 5.5 at 58) at half the input/output price”
The source says No passage could be compared.
Source cited by the article: artificialanalysis.ai/models/kimi-k3-1-pro
Couldn’t checkNumber
There are now 29 distinct providers hosting the DeepSeek V4.1 Flash 1017 variant on OpenRouter, up from 13 for DeepSeek V4 Flash in May.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “For the updated 1017 variant, there are now 29 distinct providers”
The source says No passage could be compared.
Source cited by the article: openrouter.ai/deepseek/deepseek-v4.1-flash-1017
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 262 pts, front page
Recorded for context only, checked 26 Oct 2026, 07:27 UTC. It played no part in the ranking.