OpenAI updated GPT-5.6 Sol in consumer ChatGPT with claimed 68% reduction in factual errors on financial, medical, and legal questions, while leaving the versions in Codex and ChatGPT Work unchanged. ChatGPT Plus and Pro users can now use a slider to control reasoning depth, whereas previously separate models handled this. The improvement claims lack sufficient detail for independent verification, requiring teams to test the updated model against their actual prompts.
OpenAI is removing unlimited text chat limits for free ChatGPT users and rolling out the new GPT-5.6 Luna model as the default for Free and Go tiers, while Plus and Pro users get an upgraded GPT-5.6 Sol model. Internal evaluation shows GPT-5.6 Luna reduces factual errors by 62% compared to GPT-5.5-Instant, and GPT-5.6 Sol reduces them by 68%. Free users gain unlimited text conversations alongside a new adjustable "Think" button for complex reasoning, while paid users get improved response quality and tuning controls for model reasoning depth.
OpenAI released an improved version of GPT-5.6 Sol with better accuracy and consistency, and expanded free user access to GPT-5.6 Luna for unlimited everyday conversations. The update includes performance enhancements across multiple dimensions of model capability. Free users now gain broader access to advanced model features previously restricted to paid tiers.
OpenAI released GPT-5.5, which tops objective benchmarks like the Artificial Analysis Intelligence Index with a score of 60 points, but hallucinates and confidently makes incorrect statements more often than Claude and Gemini. Pricing starts at $5 per million input tokens with xhigh reasoning mode costing $30 for output tokens. Meanwhile, major AI companies including Alphabet, Amazon, Meta, and Microsoft are building natural-gas power plants to meet AI infrastructure demands, straining their earlier net-zero climate commitments.
Anthropic launched Claude Opus 5, a vision-language model that outperforms Claude Fable 5 on many benchmarks while costing less to run, achieving the top score on Artificial Analysis' Intelligence Index. The model costs $2.03 per task on average, sits between Fable 5 ($2.75) and GPT-5.6 Sol ($1.54), and is available to Claude Max ($100-200/month) and Claude Pro ($20/month) subscribers. This update addresses longtime user complaints about Fable 5's frequent refusals, high cost, and data retention policies, making a capable model accessible to more developers at lower prices.
Every AI story that matters,
in your inbox by 8am.
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the
day in two minutes. Follow companies and topics for alerts, or get the
briefing in Slack. Free, no spam, unsubscribe anytime.
Reading TLDRocket needs no cookies, and the readership counts we rely on come from
our own cookieless analytics. Google Analytics is the exception: it sets cookies and
reports to Google, so it stays switched off until you allow it. You can change your
mind any time from “Cookie settings” in the footer.
Strictly necessary
Session security and form protection (tldrocket-session,
XSRF-TOKEN, 2 hours). The site cannot work without them,
so they need no consent.
Always on
Google Analytics 4 (_ga,
_ga_<id>, up to 2 years). Measures which
stories and sections readers use. Google acts as a third-party processor and may
store the data outside the EU. No advertising, no profiling, no data sold.
