S-tier locked
No frontier model gets S from a launch post.
Loading
Loading
Models
Season 1 week 1 tier list of frontier AI models that shipped or were the comparison set in the week of October 15, 2026. Preliminary. S-tier is locked.
As of October 15, 2026, OP or Mid's frontier-model list is preliminary. S-tier is locked. Gemini 4 Argon, GPT-6, and the Claude 5.5 family are A-tier on changelogs and published pricing, not on a hands-on bake-off. Muse Wonder is B-tier because it is three days old and Wonder Full has no API. Vendor benchmark claims are labeled and unverified.
No frontier model gets S from a launch post.
Shipped October 13, 2026. Intro price $2 / $10 per million tokens through November 30.
Catch: Number-one Vals Index and 0.7% prompt-injection claims are Google's, and we have not verified them. Price becomes $4 / $20 after November 30, 2026.
Generally available in ChatGPT. Paid tiers run Sol. Free tiers run Luna.
Catch: Ultrafast mode is on the $500 Pro plan and eligible Enterprise and Edu plans, not on Plus.
Haiku 5.5 is the price story of the week. We have not split Opus, Sonnet, and Haiku into separate ranks yet.
Catch: The newer tokenizer counts about 30% more tokens than Haiku 4.5. Over 100K tokens, Haiku 5.5 is $0.50 / $2.50.
Meta's first frontier models. Competitive on Meta's own tests. Wonder Full is locked to subscribers.
Catch: Released October 12, 2026. Wonder Full has no API. MemoryBench and the head-to-head scores are vendor claims.
Empty this week.
Empty this week.
Empty this week.
| Tier | Tool | This week | Catch | Move |
|---|---|---|---|---|
| S | S-tier locked | No frontier model gets S from a launch post. | None listed | New |
| A | Gemini 4 Argon | Shipped October 13, 2026. Intro price $2 / $10 per million tokens through November 30. | Number-one Vals Index and 0.7% prompt-injection claims are Google's, and we have not verified them. Price becomes $4 / $20 after November 30, 2026. | New |
| GPT-6 | Generally available in ChatGPT. Paid tiers run Sol. Free tiers run Luna. | Ultrafast mode is on the $500 Pro plan and eligible Enterprise and Edu plans, not on Plus. | New | |
| Claude 5.5 family | Haiku 5.5 is the price story of the week. We have not split Opus, Sonnet, and Haiku into separate ranks yet. | The newer tokenizer counts about 30% more tokens than Haiku 4.5. Over 100K tokens, Haiku 5.5 is $0.50 / $2.50. | New | |
| B | Muse Wonder | Meta's first frontier models. Competitive on Meta's own tests. Wonder Full is locked to subscribers. | Released October 12, 2026. Wonder Full has no API. MemoryBench and the head-to-head scores are vendor claims. | New |
| C | Empty this week. | |||
| D | Empty this week. | |||
| F | Empty this week. | |||
This list covers text models that shipped or were cited as the comparison set during the week of October 6–15, 2026. Ranks use official announcements and published prices. Vendor benchmark numbers are quoted as vendor claims. OP or Mid has not run a chatbot bake-off. Grok Voice 3.0 is a speech model and is not ranked here. Empty D and F rows mean no placement, not a free pass.
As of October 15, 2026, OP or Mid has locked S-tier on the frontier-model list. A launch-week benchmark claim is not an S-rank.
S-tier waits for a test we run. This week's model news is real, and it is still a press cycle. The lock is the same rule as the coding-agent list.
As of October 15, 2026, OP or Mid places Gemini 4 Argon in A-tier on its preliminary frontier-model list. The intro price is $2 / $10 per million tokens through November 30, 2026. Google's benchmark claims are labeled vendor claims and are unverified here. S-tier is locked.
Argon is live globally in the Gemini API and AI Studio, in Antigravity for Ultra subscribers, and in the Gemini app for Pro and Ultra. Cached input is 95% off during the launch window. The API offers a 1 million token unified input and output limit. Google says Argon ranks number one on the Vals Index and cuts prompt-injection attack success to 0.7% on Gray Swan's IPI benchmark. Lock experiments in before the price doubles.
As of October 15, 2026, OP or Mid places GPT-6 in A-tier on its preliminary frontier-model list. It reached every ChatGPT plan on October 7–8, 2026, with Intelligent UI. Paid tiers run Sol and free tiers run Luna. S-tier is locked.
GPT-6 with Intelligent UI went to Plus, Pro, Business, and Enterprise on October 7 and to Free and Go on October 8. Answers can include charts, buttons, forms, and mini-tools. Ultrafast mode for GPT-6.1 Sol shipped October 8 in Codex and ChatGPT Work for the Pro $500 plan and eligible Enterprise and Edu plans. Dots, OpenAI's always-on GPT-6 Astra agents, launched September 29 for Pro and Business Premium. The October 5–9 update lets you create a dot from the mobile app, and cloud browsing is faster.
As of October 15, 2026, OP or Mid places the Claude 5.5 family in A-tier on its preliminary frontier-model list. Haiku 5.5 is $0.10 / $0.50 per million tokens up to 100K tokens. The rank is not a bake-off of Opus against Haiku, and S-tier is locked.
Haiku 5.5 shipped October 7, 2026 with a 1 million token context and up to 128K output. Batch is 50% off. The same text counts as about 30% more tokens than on Haiku 4.5, so compare bills in tokens, not words. Prompts over 100K tokens cost $0.50 / $2.50. Max and Team plans include monthly API credits for your own apps and agents. Opus 5.5 and Sonnet 5.5 are part of this family rank. They are not scored separately until the test lab runs.
As of October 15, 2026, OP or Mid places Muse Wonder in B-tier on its preliminary frontier-model list. Meta claims scores comparable to Claude Fable 5.1 and GPT-6 Astra. The models were three days old, and Wonder Full has no API. S-tier is locked.
Wonder costs $4 / $20 per million tokens on the Meta Model API. Meta also says Wonder beats those comparison models on MemoryBench, its active-memory benchmark. That is Meta's test, not ours. Wonder Full stays inside Muse Code subscriptions. A new challenger with no track record yet.
Grok Voice 3.0 shipped this week in a different category: speech-to-speech at $0.08 per minute, the same price as v2.0. xAI cites Artificial Analysis for a number-one Speech-to-Speech Index score of 86.9. That figure is vendor-cited. It is not a rank on this list.