# Move routine agent work to GPT-6.1 Sol and cut cost per task by 75% > Source: > Published: 2026-09-30 11:15:58+00:00 OpenAI released GPT-6.1 Sol at $2/$10 per million tokens, and Artificial Analysis scores it 1 point below GPT-6 Astra at under a quarter of the cost per task. Agent workloads that don't need Astra can now run much cheaper. Read: OpenAI released GPT-6.1 Sol at $2/$10 per million tokens, and Artificial Analysis scores it 1 point below GPT-6 Astra at under a quarter of the cost per task. Agent workloads that don't need Astra can now run much cheaper. Read: OpenAI introduced dots, an always-on agent powered by GPT-6 Astra with its own cloud computer, which triages issues, scopes failing builds and uses Codex to return pull requests for review. Read: OpenAI launched Ultrafast for GPT-6 Astra, up to 8x faster than Standard in Codex, available to all API developers with WebSocket mode in the Responses API. Read: Anthropic's Frontier Red Team reports that open-weight GLM-5.3 produces full control-flow hijacks in 4% of binary exploitation tasks, near Claude Mythos Preview's 6%, where GLM-5.2 and Claude Opus 4.6 solved none. Discuss: Artificial Analysis measured the heaviest output token use it has recorded on Claude Sonnet 5.5, which raises cost per task about 50% above Sonnet 5 despite unchanged per-token prices. Read: Cloudflare built an LLM-driven tester that starts from known exploits and iterates on encoding and delivery using only HTTP responses, to see whether its WAF holds up against AI-style attack mutation. Read: Stripe extended Link's wallet for agents with incremental authorization for price changes at checkout, checkout obstacle guidance, spending analysis and purchase protection, as agentic purchases via Link grew 38x in a month.