The story published out there and, below it, what real token traffic from the last 7 days says about who is in it. The headline opens the original article; the model name opens its page here.
According to Anthropic, the new models show improvements in coding, reasoning and vision and generate outputs up to 30% faster than the 5.1 versions. The company charges $8 per million input tokens and $40 per million output tokens.
According to Nvidia, its board opened an independent investigation after reports that Jensen Huang was the target of a foreign intelligence effort, and the CEO admitted to infidelity in a message to employees. According to The Verge, Megaspeed executive Alice Huang was reportedly recruited by China's Ministry of State Security to spy on the company.
According to Apiiro, 1,122 public GitHub files with Claude Code settings contain passwords and other secrets. According to Ars Technica, the problem stems from the option to approve commands without asking again, which writes the command, including sensitive data, to the local settings file.
What the three paid US labs and the seven tracked Chinese labs logged in their own release notes since the previous edition. The headline opens the official note. See the lab scoreboard →
Anthropic launched Claude Fable 5.2, built for demanding reasoning and long agentic work; Mythos 5.2 offers the same only to verified organizations.
claude-fable-5-2 and claude-mythos-5-2. 1M token context window, 128k max output tokens, always-on adaptive thinking. $8 USD per MTok of input and $40 USD per MTok of output. Fable 5.2 on Claude API, AWS, Google Cloud, Bedrock and Microsoft Foundry.
Thinking blocks produced by these two models work only in the account that produced them or in an account linked to it.
Thinking blocks from Claude Fable 5.2 and Claude Mythos 5.2 work only in the account that produced them or a linked account. See Preserved thinking.
Both models support computer use through a new toolset; the earlier tool returns an error except on Amazon Bedrock.
Computer use via the computer_toolset_20260801 toolset. A request declaring computer_20251124 returns a 400 error on every platform except Amazon Bedrock, where the earlier tool still works.
In the Agents API public beta, you can change the tools of a running session, and context compaction now runs in the background by default.
v1/agents and v1/agents/sessions. Updates replace the full tool list. Set compaction.async to false for blocking compaction; background compaction may increase token usage. experimental.time_awareness true gives time reminders, disabled by default.
Tencent published the open weights of a text generation model on Hugging Face.
Qwen3.8-27B-DSpine-b16, text-generation task, open weights on Hugging Face.
From $0.26 to $0.13 per 1M tokens, blended. Input $0.1500 → $0.0060, output $0.60 → $0.50. New price since Nov 8.
What the data says. The model has 26.75% of traffic over the last 7 days, ranked 1st. DeepSeek has 40.7% of tokens and 7.5% of estimated spend this week.
See DeepSeek V4.1 Flash 1017 on Model Season →From $0.237 to $0.046 per 1M tokens, blended. Input $0.150 → $0.034, output $0.500 → $0.083. New price since Nov 8.
What the data says. The model has 6.31% of traffic over the last 7 days, ranked 3rd. Z.ai (GLM) has 9.0% of tokens and 2.1% of estimated spend this week.
See GLM 5.3 Flash on Model Season →Ranked 10th, with 2.0% of traffic over the last 7 days.
What the data says. Anonymous (stealth) has 2.1% of tokens and 0.0% of estimated spend this week.
See Crouton Alpha on Model Season →Last listed on Nov 8, 2026. Anyone calling this identifier needs to migrate.
What the data says. MiniMax has 2.9% of tokens and 0.6% of estimated spend this week. The lab's most used model is MiniMax M3.1 Flash, with 2.39% at $0.13 per 1M.
From $0.058 to $0.025 per 1M tokens, blended. Input $0.050 → $0.020, output $0.080 → $0.040. New price since Nov 8.
What the data says. No volume over the last 7 days. Its weekly peak was 1.46%.
See Llama 3.1 8B Instruct on Model Season →In the news: summaries written by AI using only what the cited sources published, reviewed before going live. The numbers in the “The data” strip come from our data, traffic from the last 7 days, never from the summary. Catalog snapshot of Nov 9, 2026, compared with the previous ones since Nov 2, 2026. Traffic through Nov 8, 2026. Price is a 75% input and 25% output blend, and only becomes news when the new value holds for two snapshots. Order follows a fixed relevance formula that weighs the type of change and the model’s share.